includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @addtogroup Parser
   6  */
   7
   8 /**
   9  * Update this version number when the ParserOutput format
  10  * changes in an incompatible way, so the parser cache
  11  * can automatically discard old data.
  12  */
  13 define( 'MW_PARSER_VERSION', '1.6.1' );
  14
  15 define( 'RLH_FOR_UPDATE', 1 );
  16
  17 # Allowed values for $mOutputType
  18 define( 'OT_HTML', 1 );
  19 define( 'OT_WIKI', 2 );
  20 define( 'OT_MSG' , 3 );
  21 define( 'OT_PREPROCESS', 4 );
  22
  23 # Flags for setFunctionHook
  24 define( 'SFH_NO_HASH', 1 );
  25
  26 # string parameter for extractTags which will cause it
  27 # to strip HTML comments in addition to regular
  28 # <XML>-style tags. This should not be anything we
  29 # may want to use in wikisyntax
  30 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  31
  32 # Constants needed for external link processing
  33 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  34 # Everything except bracket, space, or control characters
  35 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  36 # Including space, but excluding newlines
  37 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  38 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  39 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  40 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  41         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  42 define( 'EXT_IMAGE_REGEX',
  43         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  44         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  45         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  46 );
  47
  48 // State constants for the definition list colon extraction
  49 define( 'MW_COLON_STATE_TEXT', 0 );
  50 define( 'MW_COLON_STATE_TAG', 1 );
  51 define( 'MW_COLON_STATE_TAGSTART', 2 );
  52 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  53 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  54 define( 'MW_COLON_STATE_COMMENT', 5 );
  55 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  56 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  57
  58 /**
  59  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  60  * syntax, such as "[[link]]" for making links), and provides a one-way
  61  * transformation of that wiki markup it into XHTML output / markup
  62  * (which in turn the browser understands, and can display).
  63  *
  64  * <pre>
  65  * There are four main entry points into the Parser class:
  66  * parse()
  67  *   produces HTML output
  68  * preSaveTransform().
  69  *   produces altered wiki markup.
  70  * transformMsg()
  71  *   performs brace substitution on MediaWiki messages
  72  * preprocess()
  73  *   removes HTML comments and expands templates
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang, $wgContLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  84  *  $wgMaxArticleSize*
  85  *
  86  *  * only within ParserOptions
  87  * </pre>
  88  *
  89  */
  90 class Parser
  91 {
  92         const VERSION = MW_PARSER_VERSION;
  93         /**#@+
  94          * @private
  95          */
  96         # Persistent:
  97         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 103         var $mIncludeSizes, $mDefaultSort;
 104         var $mTemplates,        // cache of already loaded templates, avoids
 105                                 // multiple SQL queries for the same string
 106             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 107                                 // in this path. Used for loop detection.
 108
 109         # Temporary
 110         # These are variables reset at least once per parse regardless of $clearState
 111         var $mOptions,      // ParserOptions object
 112                 $mTitle,        // Title context, used for self-link rendering and similar things
 113                 $mOutputType,   // Output type, one of the OT_xxx constants
 114                 $ot,            // Shortcut alias, see setOutputType()
 115                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 116                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 117                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 118
 119         /**#@-*/
 120
 121         /**
 122          * Constructor
 123          *
 124          * @public
 125          */
 126         function Parser() {
 127                 $this->mTagHooks = array();
 128                 $this->mFunctionHooks = array();
 129                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 130                 $this->mFirstCall = true;
 131         }
 132
 133         /**
 134          * Do various kinds of initialisation on the first call of the parser
 135          */
 136         function firstCallInit() {
 137                 if ( !$this->mFirstCall ) {
 138                         return;
 139                 }
 140
 141                 wfProfileIn( __METHOD__ );
 142                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 143
 144                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 145
 146                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 171                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 172
 173                 if ( $wgAllowDisplayTitle ) {
 174                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 175                 }
 176                 if ( $wgAllowSlowParserFunctions ) {
 177                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 178                 }
 179
 180                 $this->initialiseVariables();
 181                 $this->mFirstCall = false;
 182                 wfProfileOut( __METHOD__ );
 183         }
 184
 185         /**
 186          * Clear Parser state
 187          *
 188          * @private
 189          */
 190         function clearState() {
 191                 wfProfileIn( __METHOD__ );
 192                 if ( $this->mFirstCall ) {
 193                         $this->firstCallInit();
 194                 }
 195                 $this->mOutput = new ParserOutput;
 196                 $this->mAutonumber = 0;
 197                 $this->mLastSection = '';
 198                 $this->mDTopen = false;
 199                 $this->mIncludeCount = array();
 200                 $this->mStripState = new StripState;
 201                 $this->mArgStack = array();
 202                 $this->mInPre = false;
 203                 $this->mInterwikiLinkHolders = array(
 204                         'texts' => array(),
 205                         'titles' => array()
 206                 );
 207                 $this->mLinkHolders = array(
 208                         'namespaces' => array(),
 209                         'dbkeys' => array(),
 210                         'queries' => array(),
 211                         'texts' => array(),
 212                         'titles' => array()
 213                 );
 214                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 215
 216                 /**
 217                  * Prefix for temporary replacement strings for the multipass parser.
 218                  * \x07 should never appear in input as it's disallowed in XML.
 219                  * Using it at the front also gives us a little extra robustness
 220                  * since it shouldn't match when butted up against identifier-like
 221                  * string constructs.
 222                  */
 223                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 224
 225                 # Clear these on every parse, bug 4549
 226                 $this->mTemplates = array();
 227                 $this->mTemplatePath = array();
 228
 229                 $this->mShowToc = true;
 230                 $this->mForceTocPosition = false;
 231                 $this->mIncludeSizes = array(
 232                         'pre-expand' => 0,
 233                         'post-expand' => 0,
 234                         'arg' => 0
 235                 );
 236                 $this->mDefaultSort = false;
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         function setOutputType( $ot ) {
 243                 $this->mOutputType = $ot;
 244                 // Shortcut alias
 245                 $this->ot = array(
 246                         'html' => $ot == OT_HTML,
 247                         'wiki' => $ot == OT_WIKI,
 248                         'msg' => $ot == OT_MSG,
 249                         'pre' => $ot == OT_PREPROCESS,
 250                 );
 251         }
 252
 253         /**
 254          * Accessor for mUniqPrefix.
 255          *
 256          * @public
 257          */
 258         function uniqPrefix() {
 259                 return $this->mUniqPrefix;
 260         }
 261
 262         /**
 263          * Convert wikitext to HTML
 264          * Do not call this function recursively.
 265          *
 266          * @param string $text Text we want to parse
 267          * @param Title &$title A title object
 268          * @param array $options
 269          * @param boolean $linestart
 270          * @param boolean $clearState
 271          * @param int $revid number to pass in {{REVISIONID}}
 272          * @return ParserOutput a ParserOutput
 273          */
 274         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 275                 /**
 276                  * First pass--just handle <nowiki> sections, pass the rest off
 277                  * to internalParse() which does all the real work.
 278                  */
 279
 280                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 281                 $fname = 'Parser::parse-' . wfGetCaller();
 282                 wfProfileIn( __METHOD__ );
 283                 wfProfileIn( $fname );
 284
 285                 if ( $clearState ) {
 286                         $this->clearState();
 287                 }
 288
 289                 $this->mOptions = $options;
 290                 $this->mTitle =& $title;
 291                 $oldRevisionId = $this->mRevisionId;
 292                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 293                 if( $revid !== null ) {
 294                         $this->mRevisionId = $revid;
 295                         $this->mRevisionTimestamp = null;
 296                 }
 297                 $this->setOutputType( OT_HTML );
 298                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 299                 $text = $this->strip( $text, $this->mStripState );
 300                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 301                 $text = $this->internalParse( $text );
 302                 $text = $this->mStripState->unstripGeneral( $text );
 303
 304                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 305                 $fixtags = array(
 306                         # french spaces, last one Guillemet-left
 307                         # only if there is something before the space
 308                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 309                         # french spaces, Guillemet-right
 310                         '/(\\302\\253) /' => '\\1&nbsp;',
 311                 );
 312                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 313
 314                 # only once and last
 315                 $text = $this->doBlockLevels( $text, $linestart );
 316
 317                 $this->replaceLinkHolders( $text );
 318
 319                 # the position of the parserConvert() call should not be changed. it
 320                 # assumes that the links are all replaced and the only thing left
 321                 # is the <nowiki> mark.
 322                 # Side-effects: this calls $this->mOutput->setTitleText()
 323                 $text = $wgContLang->parserConvert( $text, $this );
 324
 325                 $text = $this->mStripState->unstripNoWiki( $text );
 326
 327                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 328
 329                 $text = Sanitizer::normalizeCharReferences( $text );
 330
 331                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 332                         $text = Parser::tidy($text);
 333                 } else {
 334                         # attempt to sanitize at least some nesting problems
 335                         # (bug #2702 and quite a few others)
 336                         $tidyregs = array(
 337                                 # ''Something [http://www.cool.com cool''] -->
 338                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 339                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 340                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 341                                 # fix up an anchor inside another anchor, only
 342                                 # at least for a single single nested link (bug 3695)
 343                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 344                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 345                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 346                                 # contains a div, so fix it up here; replace
 347                                 # div with escaped text
 348                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 349                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 350                                 # remove empty italic or bold tag pairs, some
 351                                 # introduced by rules above
 352                                 '/<([bi])><\/\\1>/' => '',
 353                         );
 354
 355                         $text = preg_replace(
 356                                 array_keys( $tidyregs ),
 357                                 array_values( $tidyregs ),
 358                                 $text );
 359                 }
 360
 361                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 362
 363                 # Information on include size limits, for the benefit of users who try to skirt them
 364                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 365                         $max = $this->mOptions->getMaxIncludeSize();
 366                         $text .= "<!-- \n" .
 367                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 368                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 369                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 370                                 "Maximum: $max bytes\n" .
 371                                 "-->\n";
 372                 }
 373                 $this->mOutput->setText( $text );
 374                 $this->mRevisionId = $oldRevisionId;
 375                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 376                 wfProfileOut( $fname );
 377                 wfProfileOut( __METHOD__ );
 378
 379                 return $this->mOutput;
 380         }
 381
 382         /**
 383          * Recursive parser entry point that can be called from an extension tag
 384          * hook.
 385          */
 386         function recursiveTagParse( $text ) {
 387                 wfProfileIn( __METHOD__ );
 388                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 389                 $text = $this->strip( $text, $this->mStripState );
 390                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 391                 $text = $this->internalParse( $text );
 392                 wfProfileOut( __METHOD__ );
 393                 return $text;
 394         }
 395
 396         /**
 397          * Expand templates and variables in the text, producing valid, static wikitext.
 398          * Also removes comments.
 399          */
 400         function preprocess( $text, $title, $options ) {
 401                 wfProfileIn( __METHOD__ );
 402                 $this->clearState();
 403                 $this->setOutputType( OT_PREPROCESS );
 404                 $this->mOptions = $options;
 405                 $this->mTitle = $title;
 406                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 407                 $text = $this->strip( $text, $this->mStripState );
 408                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 409                 if ( $this->mOptions->getRemoveComments() ) {
 410                         $text = Sanitizer::removeHTMLcomments( $text );
 411                 }
 412                 $text = $this->replaceVariables( $text );
 413                 $text = $this->mStripState->unstripBoth( $text );
 414                 wfProfileOut( __METHOD__ );
 415                 return $text;
 416         }
 417
 418         /**
 419          * Get a random string
 420          *
 421          * @private
 422          * @static
 423          */
 424         function getRandomString() {
 425                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 426         }
 427
 428         function &getTitle() { return $this->mTitle; }
 429         function getOptions() { return $this->mOptions; }
 430
 431         function getFunctionLang() {
 432                 global $wgLang, $wgContLang;
 433                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 434         }
 435
 436         /**
 437          * Replaces all occurrences of HTML-style comments and the given tags
 438          * in the text with a random marker and returns teh next text. The output
 439          * parameter $matches will be an associative array filled with data in
 440          * the form:
 441          *   'UNIQ-xxxxx' => array(
 442          *     'element',
 443          *     'tag content',
 444          *     array( 'param' => 'x' ),
 445          *     '<element param="x">tag content</element>' ) )
 446          *
 447          * @param $elements list of element names. Comments are always extracted.
 448          * @param $text Source text string.
 449          * @param $uniq_prefix
 450          *
 451          * @private
 452          * @static
 453          */
 454         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 455                 static $n = 1;
 456                 $stripped = '';
 457                 $matches = array();
 458
 459                 $taglist = implode( '|', $elements );
 460                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 461
 462                 while ( '' != $text ) {
 463                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 464                         $stripped .= $p[0];
 465                         if( count( $p ) < 5 ) {
 466                                 break;
 467                         }
 468                         if( count( $p ) > 5 ) {
 469                                 // comment
 470                                 $element    = $p[4];
 471                                 $attributes = '';
 472                                 $close      = '';
 473                                 $inside     = $p[5];
 474                         } else {
 475                                 // tag
 476                                 $element    = $p[1];
 477                                 $attributes = $p[2];
 478                                 $close      = $p[3];
 479                                 $inside     = $p[4];
 480                         }
 481
 482                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 483                         $stripped .= $marker;
 484
 485                         if ( $close === '/>' ) {
 486                                 // Empty element tag, <tag />
 487                                 $content = null;
 488                                 $text = $inside;
 489                                 $tail = null;
 490                         } else {
 491                                 if( $element == '!--' ) {
 492                                         $end = '/(-->)/';
 493                                 } else {
 494                                         $end = "/(<\\/$element\\s*>)/i";
 495                                 }
 496                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 497                                 $content = $q[0];
 498                                 if( count( $q ) < 3 ) {
 499                                         # No end tag -- let it run out to the end of the text.
 500                                         $tail = '';
 501                                         $text = '';
 502                                 } else {
 503                                         $tail = $q[1];
 504                                         $text = $q[2];
 505                                 }
 506                         }
 507
 508                         $matches[$marker] = array( $element,
 509                                 $content,
 510                                 Sanitizer::decodeTagAttributes( $attributes ),
 511                                 "<$element$attributes$close$content$tail" );
 512                 }
 513                 return $stripped;
 514         }
 515
 516         /**
 517          * Strips and renders nowiki, pre, math, hiero
 518          * If $render is set, performs necessary rendering operations on plugins
 519          * Returns the text, and fills an array with data needed in unstrip()
 520          *
 521          * @param StripState $state
 522          *
 523          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 524          *  will be stripped in addition to other tags. This is important
 525          *  for section editing, where these comments cause confusion when
 526          *  counting the sections in the wikisource
 527          *
 528          * @param array dontstrip contains tags which should not be stripped;
 529          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 530          *
 531          * @private
 532          */
 533         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 534                 global $wgContLang;
 535                 wfProfileIn( __METHOD__ );
 536                 $render = ($this->mOutputType == OT_HTML);
 537
 538                 $uniq_prefix = $this->mUniqPrefix;
 539                 $commentState = new ReplacementArray;
 540                 $nowikiItems = array();
 541                 $generalItems = array();
 542
 543                 $elements = array_merge(
 544                         array( 'nowiki', 'gallery' ),
 545                         array_keys( $this->mTagHooks ) );
 546                 global $wgRawHtml;
 547                 if( $wgRawHtml ) {
 548                         $elements[] = 'html';
 549                 }
 550                 if( $this->mOptions->getUseTeX() ) {
 551                         $elements[] = 'math';
 552                 }
 553
 554                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 555                 foreach ( $elements AS $k => $v ) {
 556                         if ( !in_array ( $v , $dontstrip ) ) continue;
 557                         unset ( $elements[$k] );
 558                 }
 559
 560                 $matches = array();
 561                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 562
 563                 foreach( $matches as $marker => $data ) {
 564                         list( $element, $content, $params, $tag ) = $data;
 565                         if( $render ) {
 566                                 $tagName = strtolower( $element );
 567                                 wfProfileIn( __METHOD__."-render-$tagName" );
 568                                 switch( $tagName ) {
 569                                 case '!--':
 570                                         // Comment
 571                                         if( substr( $tag, -3 ) == '-->' ) {
 572                                                 $output = $tag;
 573                                         } else {
 574                                                 // Unclosed comment in input.
 575                                                 // Close it so later stripping can remove it
 576                                                 $output = "$tag-->";
 577                                         }
 578                                         break;
 579                                 case 'html':
 580                                         if( $wgRawHtml ) {
 581                                                 $output = $content;
 582                                                 break;
 583                                         }
 584                                         // Shouldn't happen otherwise. :)
 585                                 case 'nowiki':
 586                                         $output = Xml::escapeTagsOnly( $content );
 587                                         break;
 588                                 case 'math':
 589                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 590                                         break;
 591                                 case 'gallery':
 592                                         $output = $this->renderImageGallery( $content, $params );
 593                                         break;
 594                                 default:
 595                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 596                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 597                                                         array( $content, $params, $this ) );
 598                                         } else {
 599                                                 throw new MWException( "Invalid call hook $element" );
 600                                         }
 601                                 }
 602                                 wfProfileOut( __METHOD__."-render-$tagName" );
 603                         } else {
 604                                 // Just stripping tags; keep the source
 605                                 $output = $tag;
 606                         }
 607
 608                         // Unstrip the output, to support recursive strip() calls
 609                         $output = $state->unstripBoth( $output );
 610
 611                         if( !$stripcomments && $element == '!--' ) {
 612                                 $commentState->setPair( $marker, $output );
 613                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 614                                 $nowikiItems[$marker] = $output;
 615                         } else {
 616                                 $generalItems[$marker] = $output;
 617                         }
 618                 }
 619                 # Add the new items to the state
 620                 # We do this after the loop instead of during it to avoid slowing
 621                 # down the recursive unstrip
 622                 $state->nowiki->mergeArray( $nowikiItems );
 623                 $state->general->mergeArray( $generalItems );
 624
 625                 # Unstrip comments unless explicitly told otherwise.
 626                 # (The comments are always stripped prior to this point, so as to
 627                 # not invoke any extension tags / parser hooks contained within
 628                 # a comment.)
 629                 if ( !$stripcomments ) {
 630                         // Put them all back and forget them
 631                         $text = $commentState->replace( $text );
 632                 }
 633
 634                 wfProfileOut( __METHOD__ );
 635                 return $text;
 636         }
 637
 638         /**
 639          * Restores pre, math, and other extensions removed by strip()
 640          *
 641          * always call unstripNoWiki() after this one
 642          * @private
 643          * @deprecated use $this->mStripState->unstrip()
 644          */
 645         function unstrip( $text, $state ) {
 646                 return $state->unstripGeneral( $text );
 647         }
 648
 649         /**
 650          * Always call this after unstrip() to preserve the order
 651          *
 652          * @private
 653          * @deprecated use $this->mStripState->unstrip()
 654          */
 655         function unstripNoWiki( $text, $state ) {
 656                 return $state->unstripNoWiki( $text );
 657         }
 658
 659         /**
 660          * @deprecated use $this->mStripState->unstripBoth()
 661          */
 662         function unstripForHTML( $text ) {
 663                 return $this->mStripState->unstripBoth( $text );
 664         }
 665
 666         /**
 667          * Add an item to the strip state
 668          * Returns the unique tag which must be inserted into the stripped text
 669          * The tag will be replaced with the original text in unstrip()
 670          *
 671          * @private
 672          */
 673         function insertStripItem( $text, &$state ) {
 674                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 675                 $state->general->setPair( $rnd, $text );
 676                 return $rnd;
 677         }
 678
 679         /**
 680          * Interface with html tidy, used if $wgUseTidy = true.
 681          * If tidy isn't able to correct the markup, the original will be
 682          * returned in all its glory with a warning comment appended.
 683          *
 684          * Either the external tidy program or the in-process tidy extension
 685          * will be used depending on availability. Override the default
 686          * $wgTidyInternal setting to disable the internal if it's not working.
 687          *
 688          * @param string $text Hideous HTML input
 689          * @return string Corrected HTML output
 690          * @public
 691          * @static
 692          */
 693         function tidy( $text ) {
 694                 global $wgTidyInternal;
 695                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 696 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 697 '<head><title>test</title></head><body>'.$text.'</body></html>';
 698                 if( $wgTidyInternal ) {
 699                         $correctedtext = Parser::internalTidy( $wrappedtext );
 700                 } else {
 701                         $correctedtext = Parser::externalTidy( $wrappedtext );
 702                 }
 703                 if( is_null( $correctedtext ) ) {
 704                         wfDebug( "Tidy error detected!\n" );
 705                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 706                 }
 707                 return $correctedtext;
 708         }
 709
 710         /**
 711          * Spawn an external HTML tidy process and get corrected markup back from it.
 712          *
 713          * @private
 714          * @static
 715          */
 716         function externalTidy( $text ) {
 717                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 718                 $fname = 'Parser::externalTidy';
 719                 wfProfileIn( $fname );
 720
 721                 $cleansource = '';
 722                 $opts = ' -utf8';
 723
 724                 $descriptorspec = array(
 725                         0 => array('pipe', 'r'),
 726                         1 => array('pipe', 'w'),
 727                         2 => array('file', '/dev/null', 'a')  // FIXME: this line in UNIX-specific, it generates a warning on Windows, because /dev/null is not a valid Windows file.
 728                 );
 729                 $pipes = array();
 730                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 731                 if (is_resource($process)) {
 732                         // Theoretically, this style of communication could cause a deadlock
 733                         // here. If the stdout buffer fills up, then writes to stdin could
 734                         // block. This doesn't appear to happen with tidy, because tidy only
 735                         // writes to stdout after it's finished reading from stdin. Search
 736                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 737                         fwrite($pipes[0], $text);
 738                         fclose($pipes[0]);
 739                         while (!feof($pipes[1])) {
 740                                 $cleansource .= fgets($pipes[1], 1024);
 741                         }
 742                         fclose($pipes[1]);
 743                         proc_close($process);
 744                 }
 745
 746                 wfProfileOut( $fname );
 747
 748                 if( $cleansource == '' && $text != '') {
 749                         // Some kind of error happened, so we couldn't get the corrected text.
 750                         // Just give up; we'll use the source text and append a warning.
 751                         return null;
 752                 } else {
 753                         return $cleansource;
 754                 }
 755         }
 756
 757         /**
 758          * Use the HTML tidy PECL extension to use the tidy library in-process,
 759          * saving the overhead of spawning a new process. Currently written to
 760          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 761          *
 762          * 'pear install tidy' should be able to compile the extension module.
 763          *
 764          * @private
 765          * @static
 766          */
 767         function internalTidy( $text ) {
 768                 global $wgTidyConf;
 769                 $fname = 'Parser::internalTidy';
 770                 wfProfileIn( $fname );
 771
 772                 tidy_load_config( $wgTidyConf );
 773                 tidy_set_encoding( 'utf8' );
 774                 tidy_parse_string( $text );
 775                 tidy_clean_repair();
 776                 if( tidy_get_status() == 2 ) {
 777                         // 2 is magic number for fatal error
 778                         // http://www.php.net/manual/en/function.tidy-get-status.php
 779                         $cleansource = null;
 780                 } else {
 781                         $cleansource = tidy_get_output();
 782                 }
 783                 wfProfileOut( $fname );
 784                 return $cleansource;
 785         }
 786
 787         /**
 788          * parse the wiki syntax used to render tables
 789          *
 790          * @private
 791          */
 792         function doTableStuff ( $text ) {
 793                 $fname = 'Parser::doTableStuff';
 794                 wfProfileIn( $fname );
 795
 796                 $lines = explode ( "\n" , $text );
 797                 $td_history = array (); // Is currently a td tag open?
 798                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 799                 $tr_history = array (); // Is currently a tr tag open?
 800                 $tr_attributes = array (); // history of tr attributes
 801                 $has_opened_tr = array(); // Did this table open a <tr> element?
 802                 $indent_level = 0; // indent level of the table
 803                 foreach ( $lines as $key => $line )
 804                 {
 805                         $line = trim ( $line );
 806
 807                         if( $line == '' ) { // empty line, go to next line
 808                                 continue;
 809                         }
 810                         $first_character = $line{0};
 811                         $matches = array();
 812
 813                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 814                                 // First check if we are starting a new table
 815                                 $indent_level = strlen( $matches[1] );
 816
 817                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 818                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 819
 820                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 821                                 array_push ( $td_history , false );
 822                                 array_push ( $last_tag_history , '' );
 823                                 array_push ( $tr_history , false );
 824                                 array_push ( $tr_attributes , '' );
 825                                 array_push ( $has_opened_tr , false );
 826                         } else if ( count ( $td_history ) == 0 ) {
 827                                 // Don't do any of the following
 828                                 continue;
 829                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 830                                 // We are ending a table
 831                                 $line = '</table>' . substr ( $line , 2 );
 832                                 $last_tag = array_pop ( $last_tag_history );
 833
 834                                 if ( !array_pop ( $has_opened_tr ) ) {
 835                                         $line = "<tr><td></td></tr>{$line}";
 836                                 }
 837
 838                                 if ( array_pop ( $tr_history ) ) {
 839                                         $line = "</tr>{$line}";
 840                                 }
 841
 842                                 if ( array_pop ( $td_history ) ) {
 843                                         $line = "</{$last_tag}>{$line}";
 844                                 }
 845                                 array_pop ( $tr_attributes );
 846                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 847                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 848                                 // Now we have a table row
 849                                 $line = preg_replace( '#^\|-+#', '', $line );
 850
 851                                 // Whats after the tag is now only attributes
 852                                 $attributes = $this->mStripState->unstripBoth( $line );
 853                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 854                                 array_pop ( $tr_attributes );
 855                                 array_push ( $tr_attributes , $attributes );
 856
 857                                 $line = '';
 858                                 $last_tag = array_pop ( $last_tag_history );
 859                                 array_pop ( $has_opened_tr );
 860                                 array_push ( $has_opened_tr , true );
 861
 862                                 if ( array_pop ( $tr_history ) ) {
 863                                         $line = '</tr>';
 864                                 }
 865
 866                                 if ( array_pop ( $td_history ) ) {
 867                                         $line = "</{$last_tag}>{$line}";
 868                                 }
 869
 870                                 $lines[$key] = $line;
 871                                 array_push ( $tr_history , false );
 872                                 array_push ( $td_history , false );
 873                                 array_push ( $last_tag_history , '' );
 874                         }
 875                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 876                                 // This might be cell elements, td, th or captions
 877                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 878                                         $first_character = '+';
 879                                         $line = substr ( $line , 1 );
 880                                 }
 881
 882                                 $line = substr ( $line , 1 );
 883
 884                                 if ( $first_character == '!' ) {
 885                                         $line = str_replace ( '!!' , '||' , $line );
 886                                 }
 887
 888                                 // Split up multiple cells on the same line.
 889                                 // FIXME : This can result in improper nesting of tags processed
 890                                 // by earlier parser steps, but should avoid splitting up eg
 891                                 // attribute values containing literal "||".
 892                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 893
 894                                 $lines[$key] = '';
 895
 896                                 // Loop through each table cell
 897                                 foreach ( $cells as $cell )
 898                                 {
 899                                         $previous = '';
 900                                         if ( $first_character != '+' )
 901                                         {
 902                                                 $tr_after = array_pop ( $tr_attributes );
 903                                                 if ( !array_pop ( $tr_history ) ) {
 904                                                         $previous = "<tr{$tr_after}>\n";
 905                                                 }
 906                                                 array_push ( $tr_history , true );
 907                                                 array_push ( $tr_attributes , '' );
 908                                                 array_pop ( $has_opened_tr );
 909                                                 array_push ( $has_opened_tr , true );
 910                                         }
 911
 912                                         $last_tag = array_pop ( $last_tag_history );
 913
 914                                         if ( array_pop ( $td_history ) ) {
 915                                                 $previous = "</{$last_tag}>{$previous}";
 916                                         }
 917
 918                                         if ( $first_character == '|' ) {
 919                                                 $last_tag = 'td';
 920                                         } else if ( $first_character == '!' ) {
 921                                                 $last_tag = 'th';
 922                                         } else if ( $first_character == '+' ) {
 923                                                 $last_tag = 'caption';
 924                                         } else {
 925                                                 $last_tag = '';
 926                                         }
 927
 928                                         array_push ( $last_tag_history , $last_tag );
 929
 930                                         // A cell could contain both parameters and data
 931                                         $cell_data = explode ( '|' , $cell , 2 );
 932
 933                                         // Bug 553: Note that a '|' inside an invalid link should not
 934                                         // be mistaken as delimiting cell parameters
 935                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 936                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 937                                         } else if ( count ( $cell_data ) == 1 )
 938                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 939                                         else {
 940                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 941                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 942                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 943                                         }
 944
 945                                         $lines[$key] .= $cell;
 946                                         array_push ( $td_history , true );
 947                                 }
 948                         }
 949                 }
 950
 951                 // Closing open td, tr && table
 952                 while ( count ( $td_history ) > 0 )
 953                 {
 954                         if ( array_pop ( $td_history ) ) {
 955                                 $lines[] = '</td>' ;
 956                         }
 957                         if ( array_pop ( $tr_history ) ) {
 958                                 $lines[] = '</tr>' ;
 959                         }
 960                         if ( !array_pop ( $has_opened_tr ) ) {
 961                                 $lines[] = "<tr><td></td></tr>" ;
 962                         }
 963
 964                         $lines[] = '</table>' ;
 965                 }
 966
 967                 $output = implode ( "\n" , $lines ) ;
 968
 969                 // special case: don't return empty table
 970                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 971                         $output = '';
 972                 }
 973
 974                 wfProfileOut( $fname );
 975
 976                 return $output;
 977         }
 978
 979         /**
 980          * Helper function for parse() that transforms wiki markup into
 981          * HTML. Only called for $mOutputType == OT_HTML.
 982          *
 983          * @private
 984          */
 985         function internalParse( $text ) {
 986                 $args = array();
 987                 $isMain = true;
 988                 $fname = 'Parser::internalParse';
 989                 wfProfileIn( $fname );
 990
 991                 # Hook to suspend the parser in this state
 992                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 993                         wfProfileOut( $fname );
 994                         return $text ;
 995                 }
 996
 997                 # Remove <noinclude> tags and <includeonly> sections
 998                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 999                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
1000                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1001
1002                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
1003
1004                 $text = $this->replaceVariables( $text, $args );
1005                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text ) );
1006
1007                 // Tables need to come after variable replacement for things to work
1008                 // properly; putting them before other transformations should keep
1009                 // exciting things like link expansions from showing up in surprising
1010                 // places.
1011                 $text = $this->doTableStuff( $text );
1012
1013                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1014
1015                 $text = $this->stripToc( $text );
1016                 $this->stripNoGallery( $text );
1017                 $text = $this->doHeadings( $text );
1018                 if($this->mOptions->getUseDynamicDates()) {
1019                         $df =& DateFormatter::getInstance();
1020                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1021                 }
1022                 $text = $this->doAllQuotes( $text );
1023                 $text = $this->replaceInternalLinks( $text );
1024                 $text = $this->replaceExternalLinks( $text );
1025
1026                 # replaceInternalLinks may sometimes leave behind
1027                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1028                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1029
1030                 $text = $this->doMagicLinks( $text );
1031                 $text = $this->formatHeadings( $text, $isMain );
1032
1033                 wfProfileOut( $fname );
1034                 return $text;
1035         }
1036
1037         /**
1038          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1039          * magic external links.
1040          *
1041          * @private
1042          */
1043         function &doMagicLinks( &$text ) {
1044                 wfProfileIn( __METHOD__ );
1045                 $text = preg_replace_callback(
1046                         '!(?:                           # Start cases
1047                             <a.*?</a> |                 # Skip link text
1048                             <.*?> |                     # Skip stuff inside HTML elements
1049                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1050                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1051                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1052                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1053                                      [0-9Xx]                 # check digit
1054                                    \b)
1055                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1056                 wfProfileOut( __METHOD__ );
1057                 return $text;
1058         }
1059
1060         function magicLinkCallback( $m ) {
1061                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1062                         # Skip HTML element
1063                         return $m[0];
1064                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1065                         $isbn = $m[2];
1066                         $num = strtr( $isbn, array(
1067                                 '-' => '',
1068                                 ' ' => '',
1069                                 'x' => 'X',
1070                         ));
1071                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1072                         $text = '<a href="' .
1073                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1074                                 "\" class=\"internal\">ISBN $isbn</a>";
1075                 } else {
1076                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1077                                 $keyword = 'RFC';
1078                                 $urlmsg = 'rfcurl';
1079                                 $id = $m[1];
1080                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1081                                 $keyword = 'PMID';
1082                                 $urlmsg = 'pubmedurl';
1083                                 $id = $m[1];
1084                         } else {
1085                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1086                                         substr($m[0], 0, 20 ) . '"' );
1087                         }
1088
1089                         $url = wfMsg( $urlmsg, $id);
1090                         $sk = $this->mOptions->getSkin();
1091                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1092                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1093                 }
1094                 return $text;
1095         }
1096
1097         /**
1098          * Parse headers and return html
1099          *
1100          * @private
1101          */
1102         function doHeadings( $text ) {
1103                 $fname = 'Parser::doHeadings';
1104                 wfProfileIn( $fname );
1105                 for ( $i = 6; $i >= 1; --$i ) {
1106                         $h = str_repeat( '=', $i );
1107                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1108                           "<h{$i}>\\1</h{$i}>\\2", $text );
1109                 }
1110                 wfProfileOut( $fname );
1111                 return $text;
1112         }
1113
1114         /**
1115          * Replace single quotes with HTML markup
1116          * @private
1117          * @return string the altered text
1118          */
1119         function doAllQuotes( $text ) {
1120                 $fname = 'Parser::doAllQuotes';
1121                 wfProfileIn( $fname );
1122                 $outtext = '';
1123                 $lines = explode( "\n", $text );
1124                 foreach ( $lines as $line ) {
1125                         $outtext .= $this->doQuotes ( $line ) . "\n";
1126                 }
1127                 $outtext = substr($outtext, 0,-1);
1128                 wfProfileOut( $fname );
1129                 return $outtext;
1130         }
1131
1132         /**
1133          * Helper function for doAllQuotes()
1134          * @private
1135          */
1136         function doQuotes( $text ) {
1137                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1138                 if ( count( $arr ) == 1 )
1139                         return $text;
1140                 else
1141                 {
1142                         # First, do some preliminary work. This may shift some apostrophes from
1143                         # being mark-up to being text. It also counts the number of occurrences
1144                         # of bold and italics mark-ups.
1145                         $i = 0;
1146                         $numbold = 0;
1147                         $numitalics = 0;
1148                         foreach ( $arr as $r )
1149                         {
1150                                 if ( ( $i % 2 ) == 1 )
1151                                 {
1152                                         # If there are ever four apostrophes, assume the first is supposed to
1153                                         # be text, and the remaining three constitute mark-up for bold text.
1154                                         if ( strlen( $arr[$i] ) == 4 )
1155                                         {
1156                                                 $arr[$i-1] .= "'";
1157                                                 $arr[$i] = "'''";
1158                                         }
1159                                         # If there are more than 5 apostrophes in a row, assume they're all
1160                                         # text except for the last 5.
1161                                         else if ( strlen( $arr[$i] ) > 5 )
1162                                         {
1163                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1164                                                 $arr[$i] = "'''''";
1165                                         }
1166                                         # Count the number of occurrences of bold and italics mark-ups.
1167                                         # We are not counting sequences of five apostrophes.
1168                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1169                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1170                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1171                                 }
1172                                 $i++;
1173                         }
1174
1175                         # If there is an odd number of both bold and italics, it is likely
1176                         # that one of the bold ones was meant to be an apostrophe followed
1177                         # by italics. Which one we cannot know for certain, but it is more
1178                         # likely to be one that has a single-letter word before it.
1179                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1180                         {
1181                                 $i = 0;
1182                                 $firstsingleletterword = -1;
1183                                 $firstmultiletterword = -1;
1184                                 $firstspace = -1;
1185                                 foreach ( $arr as $r )
1186                                 {
1187                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1188                                         {
1189                                                 $x1 = substr ($arr[$i-1], -1);
1190                                                 $x2 = substr ($arr[$i-1], -2, 1);
1191                                                 if ($x1 == ' ') {
1192                                                         if ($firstspace == -1) $firstspace = $i;
1193                                                 } else if ($x2 == ' ') {
1194                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1195                                                 } else {
1196                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1197                                                 }
1198                                         }
1199                                         $i++;
1200                                 }
1201
1202                                 # If there is a single-letter word, use it!
1203                                 if ($firstsingleletterword > -1)
1204                                 {
1205                                         $arr [ $firstsingleletterword ] = "''";
1206                                         $arr [ $firstsingleletterword-1 ] .= "'";
1207                                 }
1208                                 # If not, but there's a multi-letter word, use that one.
1209                                 else if ($firstmultiletterword > -1)
1210                                 {
1211                                         $arr [ $firstmultiletterword ] = "''";
1212                                         $arr [ $firstmultiletterword-1 ] .= "'";
1213                                 }
1214                                 # ... otherwise use the first one that has neither.
1215                                 # (notice that it is possible for all three to be -1 if, for example,
1216                                 # there is only one pentuple-apostrophe in the line)
1217                                 else if ($firstspace > -1)
1218                                 {
1219                                         $arr [ $firstspace ] = "''";
1220                                         $arr [ $firstspace-1 ] .= "'";
1221                                 }
1222                         }
1223
1224                         # Now let's actually convert our apostrophic mush to HTML!
1225                         $output = '';
1226                         $buffer = '';
1227                         $state = '';
1228                         $i = 0;
1229                         foreach ($arr as $r)
1230                         {
1231                                 if (($i % 2) == 0)
1232                                 {
1233                                         if ($state == 'both')
1234                                                 $buffer .= $r;
1235                                         else
1236                                                 $output .= $r;
1237                                 }
1238                                 else
1239                                 {
1240                                         if (strlen ($r) == 2)
1241                                         {
1242                                                 if ($state == 'i')
1243                                                 { $output .= '</i>'; $state = ''; }
1244                                                 else if ($state == 'bi')
1245                                                 { $output .= '</i>'; $state = 'b'; }
1246                                                 else if ($state == 'ib')
1247                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1248                                                 else if ($state == 'both')
1249                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1250                                                 else # $state can be 'b' or ''
1251                                                 { $output .= '<i>'; $state .= 'i'; }
1252                                         }
1253                                         else if (strlen ($r) == 3)
1254                                         {
1255                                                 if ($state == 'b')
1256                                                 { $output .= '</b>'; $state = ''; }
1257                                                 else if ($state == 'bi')
1258                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1259                                                 else if ($state == 'ib')
1260                                                 { $output .= '</b>'; $state = 'i'; }
1261                                                 else if ($state == 'both')
1262                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1263                                                 else # $state can be 'i' or ''
1264                                                 { $output .= '<b>'; $state .= 'b'; }
1265                                         }
1266                                         else if (strlen ($r) == 5)
1267                                         {
1268                                                 if ($state == 'b')
1269                                                 { $output .= '</b><i>'; $state = 'i'; }
1270                                                 else if ($state == 'i')
1271                                                 { $output .= '</i><b>'; $state = 'b'; }
1272                                                 else if ($state == 'bi')
1273                                                 { $output .= '</i></b>'; $state = ''; }
1274                                                 else if ($state == 'ib')
1275                                                 { $output .= '</b></i>'; $state = ''; }
1276                                                 else if ($state == 'both')
1277                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1278                                                 else # ($state == '')
1279                                                 { $buffer = ''; $state = 'both'; }
1280                                         }
1281                                 }
1282                                 $i++;
1283                         }
1284                         # Now close all remaining tags.  Notice that the order is important.
1285                         if ($state == 'b' || $state == 'ib')
1286                                 $output .= '</b>';
1287                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1288                                 $output .= '</i>';
1289                         if ($state == 'bi')
1290                                 $output .= '</b>';
1291                         # There might be lonely ''''', so make sure we have a buffer
1292                         if ($state == 'both' && $buffer)
1293                                 $output .= '<b><i>'.$buffer.'</i></b>';
1294                         return $output;
1295                 }
1296         }
1297
1298         /**
1299          * Replace external links
1300          *
1301          * Note: this is all very hackish and the order of execution matters a lot.
1302          * Make sure to run maintenance/parserTests.php if you change this code.
1303          *
1304          * @private
1305          */
1306         function replaceExternalLinks( $text ) {
1307                 global $wgContLang;
1308                 $fname = 'Parser::replaceExternalLinks';
1309                 wfProfileIn( $fname );
1310
1311                 $sk = $this->mOptions->getSkin();
1312
1313                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1314
1315                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1316
1317                 $i = 0;
1318                 while ( $i<count( $bits ) ) {
1319                         $url = $bits[$i++];
1320                         $protocol = $bits[$i++];
1321                         $text = $bits[$i++];
1322                         $trail = $bits[$i++];
1323
1324                         # The characters '<' and '>' (which were escaped by
1325                         # removeHTMLtags()) should not be included in
1326                         # URLs, per RFC 2396.
1327                         $m2 = array();
1328                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1329                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1330                                 $url = substr($url, 0, $m2[0][1]);
1331                         }
1332
1333                         # If the link text is an image URL, replace it with an <img> tag
1334                         # This happened by accident in the original parser, but some people used it extensively
1335                         $img = $this->maybeMakeExternalImage( $text );
1336                         if ( $img !== false ) {
1337                                 $text = $img;
1338                         }
1339
1340                         $dtrail = '';
1341
1342                         # Set linktype for CSS - if URL==text, link is essentially free
1343                         $linktype = ($text == $url) ? 'free' : 'text';
1344
1345                         # No link text, e.g. [http://domain.tld/some.link]
1346                         if ( $text == '' ) {
1347                                 # Autonumber if allowed. See bug #5918
1348                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1349                                         $text = '[' . ++$this->mAutonumber . ']';
1350                                         $linktype = 'autonumber';
1351                                 } else {
1352                                         # Otherwise just use the URL
1353                                         $text = htmlspecialchars( $url );
1354                                         $linktype = 'free';
1355                                 }
1356                         } else {
1357                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1358                                 # Check for trail
1359                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1360                         }
1361
1362                         $text = $wgContLang->markNoConversion($text);
1363
1364                         $url = Sanitizer::cleanUrl( $url );
1365
1366                         # Process the trail (i.e. everything after this link up until start of the next link),
1367                         # replacing any non-bracketed links
1368                         $trail = $this->replaceFreeExternalLinks( $trail );
1369
1370                         # Use the encoded URL
1371                         # This means that users can paste URLs directly into the text
1372                         # Funny characters like &ouml; aren't valid in URLs anyway
1373                         # This was changed in August 2004
1374                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1375
1376                         # Register link in the output object.
1377                         # Replace unnecessary URL escape codes with the referenced character
1378                         # This prevents spammers from hiding links from the filters
1379                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1380                         $this->mOutput->addExternalLink( $pasteurized );
1381                 }
1382
1383                 wfProfileOut( $fname );
1384                 return $s;
1385         }
1386
1387         /**
1388          * Replace anything that looks like a URL with a link
1389          * @private
1390          */
1391         function replaceFreeExternalLinks( $text ) {
1392                 global $wgContLang;
1393                 $fname = 'Parser::replaceFreeExternalLinks';
1394                 wfProfileIn( $fname );
1395
1396                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1397                 $s = array_shift( $bits );
1398                 $i = 0;
1399
1400                 $sk = $this->mOptions->getSkin();
1401
1402                 while ( $i < count( $bits ) ){
1403                         $protocol = $bits[$i++];
1404                         $remainder = $bits[$i++];
1405
1406                         $m = array();
1407                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1408                                 # Found some characters after the protocol that look promising
1409                                 $url = $protocol . $m[1];
1410                                 $trail = $m[2];
1411
1412                                 # special case: handle urls as url args:
1413                                 # http://www.example.com/foo?=http://www.example.com/bar
1414                                 if(strlen($trail) == 0 &&
1415                                         isset($bits[$i]) &&
1416                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1417                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1418                                 {
1419                                         # add protocol, arg
1420                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1421                                         $i += 2;
1422                                         $trail = $m[2];
1423                                 }
1424
1425                                 # The characters '<' and '>' (which were escaped by
1426                                 # removeHTMLtags()) should not be included in
1427                                 # URLs, per RFC 2396.
1428                                 $m2 = array();
1429                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1430                                         $trail = substr($url, $m2[0][1]) . $trail;
1431                                         $url = substr($url, 0, $m2[0][1]);
1432                                 }
1433
1434                                 # Move trailing punctuation to $trail
1435                                 $sep = ',;\.:!?';
1436                                 # If there is no left bracket, then consider right brackets fair game too
1437                                 if ( strpos( $url, '(' ) === false ) {
1438                                         $sep .= ')';
1439                                 }
1440
1441                                 $numSepChars = strspn( strrev( $url ), $sep );
1442                                 if ( $numSepChars ) {
1443                                         $trail = substr( $url, -$numSepChars ) . $trail;
1444                                         $url = substr( $url, 0, -$numSepChars );
1445                                 }
1446
1447                                 $url = Sanitizer::cleanUrl( $url );
1448
1449                                 # Is this an external image?
1450                                 $text = $this->maybeMakeExternalImage( $url );
1451                                 if ( $text === false ) {
1452                                         # Not an image, make a link
1453                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1454                                         # Register it in the output object...
1455                                         # Replace unnecessary URL escape codes with their equivalent characters
1456                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1457                                         $this->mOutput->addExternalLink( $pasteurized );
1458                                 }
1459                                 $s .= $text . $trail;
1460                         } else {
1461                                 $s .= $protocol . $remainder;
1462                         }
1463                 }
1464                 wfProfileOut( $fname );
1465                 return $s;
1466         }
1467
1468         /**
1469          * Replace unusual URL escape codes with their equivalent characters
1470          * @param string
1471          * @return string
1472          * @static
1473          * @todo  This can merge genuinely required bits in the path or query string,
1474          *        breaking legit URLs. A proper fix would treat the various parts of
1475          *        the URL differently; as a workaround, just use the output for
1476          *        statistical records, not for actual linking/output.
1477          */
1478         static function replaceUnusualEscapes( $url ) {
1479                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1480                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1481         }
1482
1483         /**
1484          * Callback function used in replaceUnusualEscapes().
1485          * Replaces unusual URL escape codes with their equivalent character
1486          * @static
1487          * @private
1488          */
1489         private static function replaceUnusualEscapesCallback( $matches ) {
1490                 $char = urldecode( $matches[0] );
1491                 $ord = ord( $char );
1492                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1493                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1494                         // No, shouldn't be escaped
1495                         return $char;
1496                 } else {
1497                         // Yes, leave it escaped
1498                         return $matches[0];
1499                 }
1500         }
1501
1502         /**
1503          * make an image if it's allowed, either through the global
1504          * option or through the exception
1505          * @private
1506          */
1507         function maybeMakeExternalImage( $url ) {
1508                 $sk = $this->mOptions->getSkin();
1509                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1510                 $imagesexception = !empty($imagesfrom);
1511                 $text = false;
1512                 if ( $this->mOptions->getAllowExternalImages()
1513                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1514                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1515                                 # Image found
1516                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1517                         }
1518                 }
1519                 return $text;
1520         }
1521
1522         /**
1523          * Process [[ ]] wikilinks
1524          *
1525          * @private
1526          */
1527         function replaceInternalLinks( $s ) {
1528                 global $wgContLang;
1529                 static $fname = 'Parser::replaceInternalLinks' ;
1530
1531                 wfProfileIn( $fname );
1532
1533                 wfProfileIn( $fname.'-setup' );
1534                 static $tc = FALSE;
1535                 # the % is needed to support urlencoded titles as well
1536                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1537
1538                 $sk = $this->mOptions->getSkin();
1539
1540                 #split the entire text string on occurences of [[
1541                 $a = explode( '[[', ' ' . $s );
1542                 #get the first element (all text up to first [[), and remove the space we added
1543                 $s = array_shift( $a );
1544                 $s = substr( $s, 1 );
1545
1546                 # Match a link having the form [[namespace:link|alternate]]trail
1547                 static $e1 = FALSE;
1548                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1549                 # Match cases where there is no "]]", which might still be images
1550                 static $e1_img = FALSE;
1551                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1552                 # Match the end of a line for a word that's not followed by whitespace,
1553                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1554                 $e2 = wfMsgForContent( 'linkprefix' );
1555
1556                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1557                 if( is_null( $this->mTitle ) ) {
1558                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1559                 }
1560                 $nottalk = !$this->mTitle->isTalkPage();
1561
1562                 if ( $useLinkPrefixExtension ) {
1563                         $m = array();
1564                         if ( preg_match( $e2, $s, $m ) ) {
1565                                 $first_prefix = $m[2];
1566                         } else {
1567                                 $first_prefix = false;
1568                         }
1569                 } else {
1570                         $prefix = '';
1571                 }
1572
1573                 if($wgContLang->hasVariants()) {
1574                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1575                 } else {
1576                         $selflink = array($this->mTitle->getPrefixedText());
1577                 }
1578                 $useSubpages = $this->areSubpagesAllowed();
1579                 wfProfileOut( $fname.'-setup' );
1580
1581                 # Loop for each link
1582                 for ($k = 0; isset( $a[$k] ); $k++) {
1583                         $line = $a[$k];
1584                         if ( $useLinkPrefixExtension ) {
1585                                 wfProfileIn( $fname.'-prefixhandling' );
1586                                 if ( preg_match( $e2, $s, $m ) ) {
1587                                         $prefix = $m[2];
1588                                         $s = $m[1];
1589                                 } else {
1590                                         $prefix='';
1591                                 }
1592                                 # first link
1593                                 if($first_prefix) {
1594                                         $prefix = $first_prefix;
1595                                         $first_prefix = false;
1596                                 }
1597                                 wfProfileOut( $fname.'-prefixhandling' );
1598                         }
1599
1600                         $might_be_img = false;
1601
1602                         wfProfileIn( "$fname-e1" );
1603                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1604                                 $text = $m[2];
1605                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1606                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1607                                 # the real problem is with the $e1 regex
1608                                 # See bug 1300.
1609                                 #
1610                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1611                                 # and no image is in sight. See bug 2095.
1612                                 #
1613                                 if( $text !== '' &&
1614                                         substr( $m[3], 0, 1 ) === ']' &&
1615                                         strpos($text, '[') !== false
1616                                 )
1617                                 {
1618                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1619                                         $m[3] = substr( $m[3], 1 );
1620                                 }
1621                                 # fix up urlencoded title texts
1622                                 if( strpos( $m[1], '%' ) !== false ) {
1623                                         # Should anchors '#' also be rejected?
1624                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1625                                 }
1626                                 $trail = $m[3];
1627                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1628                                 $might_be_img = true;
1629                                 $text = $m[2];
1630                                 if ( strpos( $m[1], '%' ) !== false ) {
1631                                         $m[1] = urldecode($m[1]);
1632                                 }
1633                                 $trail = "";
1634                         } else { # Invalid form; output directly
1635                                 $s .= $prefix . '[[' . $line ;
1636                                 wfProfileOut( "$fname-e1" );
1637                                 continue;
1638                         }
1639                         wfProfileOut( "$fname-e1" );
1640                         wfProfileIn( "$fname-misc" );
1641
1642                         # Don't allow internal links to pages containing
1643                         # PROTO: where PROTO is a valid URL protocol; these
1644                         # should be external links.
1645                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1646                                 $s .= $prefix . '[[' . $line ;
1647                                 continue;
1648                         }
1649
1650                         # Make subpage if necessary
1651                         if( $useSubpages ) {
1652                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1653                         } else {
1654                                 $link = $m[1];
1655                         }
1656
1657                         $noforce = (substr($m[1], 0, 1) != ':');
1658                         if (!$noforce) {
1659                                 # Strip off leading ':'
1660                                 $link = substr($link, 1);
1661                         }
1662
1663                         wfProfileOut( "$fname-misc" );
1664                         wfProfileIn( "$fname-title" );
1665                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1666                         if( !$nt ) {
1667                                 $s .= $prefix . '[[' . $line;
1668                                 wfProfileOut( "$fname-title" );
1669                                 continue;
1670                         }
1671
1672                         $ns = $nt->getNamespace();
1673                         $iw = $nt->getInterWiki();
1674                         wfProfileOut( "$fname-title" );
1675
1676                         if ($might_be_img) { # if this is actually an invalid link
1677                                 wfProfileIn( "$fname-might_be_img" );
1678                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1679                                         $found = false;
1680                                         while (isset ($a[$k+1]) ) {
1681                                                 #look at the next 'line' to see if we can close it there
1682                                                 $spliced = array_splice( $a, $k + 1, 1 );
1683                                                 $next_line = array_shift( $spliced );
1684                                                 $m = explode( ']]', $next_line, 3 );
1685                                                 if ( count( $m ) == 3 ) {
1686                                                         # the first ]] closes the inner link, the second the image
1687                                                         $found = true;
1688                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1689                                                         $trail = $m[2];
1690                                                         break;
1691                                                 } elseif ( count( $m ) == 2 ) {
1692                                                         #if there's exactly one ]] that's fine, we'll keep looking
1693                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1694                                                 } else {
1695                                                         #if $next_line is invalid too, we need look no further
1696                                                         $text .= '[[' . $next_line;
1697                                                         break;
1698                                                 }
1699                                         }
1700                                         if ( !$found ) {
1701                                                 # we couldn't find the end of this imageLink, so output it raw
1702                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1703                                                 $text = $this->replaceInternalLinks($text);
1704                                                 $s .= "{$prefix}[[$link|$text";
1705                                                 # note: no $trail, because without an end, there *is* no trail
1706                                                 wfProfileOut( "$fname-might_be_img" );
1707                                                 continue;
1708                                         }
1709                                 } else { #it's not an image, so output it raw
1710                                         $s .= "{$prefix}[[$link|$text";
1711                                         # note: no $trail, because without an end, there *is* no trail
1712                                         wfProfileOut( "$fname-might_be_img" );
1713                                         continue;
1714                                 }
1715                                 wfProfileOut( "$fname-might_be_img" );
1716                         }
1717
1718                         $wasblank = ( '' == $text );
1719                         if( $wasblank ) $text = $link;
1720
1721                         # Link not escaped by : , create the various objects
1722                         if( $noforce ) {
1723
1724                                 # Interwikis
1725                                 wfProfileIn( "$fname-interwiki" );
1726                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1727                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1728                                         $s = rtrim($s . $prefix);
1729                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1730                                         wfProfileOut( "$fname-interwiki" );
1731                                         continue;
1732                                 }
1733                                 wfProfileOut( "$fname-interwiki" );
1734
1735                                 if ( $ns == NS_IMAGE ) {
1736                                         wfProfileIn( "$fname-image" );
1737                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1738                                                 # recursively parse links inside the image caption
1739                                                 # actually, this will parse them in any other parameters, too,
1740                                                 # but it might be hard to fix that, and it doesn't matter ATM
1741                                                 $text = $this->replaceExternalLinks($text);
1742                                                 $text = $this->replaceInternalLinks($text);
1743
1744                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1745                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1746                                                 $this->mOutput->addImage( $nt->getDBkey() );
1747
1748                                                 wfProfileOut( "$fname-image" );
1749                                                 continue;
1750                                         } else {
1751                                                 # We still need to record the image's presence on the page
1752                                                 $this->mOutput->addImage( $nt->getDBkey() );
1753                                         }
1754                                         wfProfileOut( "$fname-image" );
1755
1756                                 }
1757
1758                                 if ( $ns == NS_CATEGORY ) {
1759                                         wfProfileIn( "$fname-category" );
1760                                         $s = rtrim($s . "\n"); # bug 87
1761
1762                                         if ( $wasblank ) {
1763                                                 $sortkey = $this->getDefaultSort();
1764                                         } else {
1765                                                 $sortkey = $text;
1766                                         }
1767                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1768                                         $sortkey = str_replace( "\n", '', $sortkey );
1769                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1770                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1771
1772                                         /**
1773                                          * Strip the whitespace Category links produce, see bug 87
1774                                          * @todo We might want to use trim($tmp, "\n") here.
1775                                          */
1776                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1777
1778                                         wfProfileOut( "$fname-category" );
1779                                         continue;
1780                                 }
1781                         }
1782
1783                         # Self-link checking
1784                         if( $nt->getFragment() === '' ) {
1785                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1786                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1787                                         continue;
1788                                 }
1789                         }
1790
1791                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1792                         if( $ns == NS_MEDIA ) {
1793                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1794                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1795                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1796                                 $this->mOutput->addImage( $nt->getDBkey() );
1797                                 continue;
1798                         } elseif( $ns == NS_SPECIAL ) {
1799                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1800                                 continue;
1801                         } elseif( $ns == NS_IMAGE ) {
1802                                 $img = new Image( $nt );
1803                                 if( $img->exists() ) {
1804                                         // Force a blue link if the file exists; may be a remote
1805                                         // upload on the shared repository, and we want to see its
1806                                         // auto-generated page.
1807                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1808                                         $this->mOutput->addLink( $nt );
1809                                         continue;
1810                                 }
1811                         }
1812                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1813                 }
1814                 wfProfileOut( $fname );
1815                 return $s;
1816         }
1817
1818         /**
1819          * Make a link placeholder. The text returned can be later resolved to a real link with
1820          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1821          * parsing of interwiki links, and secondly to allow all existence checks and
1822          * article length checks (for stub links) to be bundled into a single query.
1823          *
1824          */
1825         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1826                 wfProfileIn( __METHOD__ );
1827                 if ( ! is_object($nt) ) {
1828                         # Fail gracefully
1829                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1830                 } else {
1831                         # Separate the link trail from the rest of the link
1832                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1833
1834                         if ( $nt->isExternal() ) {
1835                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1836                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1837                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1838                         } else {
1839                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1840                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1841                                 $this->mLinkHolders['queries'][] = $query;
1842                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1843                                 $this->mLinkHolders['titles'][] = $nt;
1844
1845                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1846                         }
1847                 }
1848                 wfProfileOut( __METHOD__ );
1849                 return $retVal;
1850         }
1851
1852         /**
1853          * Render a forced-blue link inline; protect against double expansion of
1854          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1855          * Since this little disaster has to split off the trail text to avoid
1856          * breaking URLs in the following text without breaking trails on the
1857          * wiki links, it's been made into a horrible function.
1858          *
1859          * @param Title $nt
1860          * @param string $text
1861          * @param string $query
1862          * @param string $trail
1863          * @param string $prefix
1864          * @return string HTML-wikitext mix oh yuck
1865          */
1866         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1867                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1868                 $sk = $this->mOptions->getSkin();
1869                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1870                 return $this->armorLinks( $link ) . $trail;
1871         }
1872
1873         /**
1874          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1875          * going to go through further parsing steps before inline URL expansion.
1876          *
1877          * In particular this is important when using action=render, which causes
1878          * full URLs to be included.
1879          *
1880          * Oh man I hate our multi-layer parser!
1881          *
1882          * @param string more-or-less HTML
1883          * @return string less-or-more HTML with NOPARSE bits
1884          */
1885         function armorLinks( $text ) {
1886                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1887                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1888         }
1889
1890         /**
1891          * Return true if subpage links should be expanded on this page.
1892          * @return bool
1893          */
1894         function areSubpagesAllowed() {
1895                 # Some namespaces don't allow subpages
1896                 global $wgNamespacesWithSubpages;
1897                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1898         }
1899
1900         /**
1901          * Handle link to subpage if necessary
1902          * @param string $target the source of the link
1903          * @param string &$text the link text, modified as necessary
1904          * @return string the full name of the link
1905          * @private
1906          */
1907         function maybeDoSubpageLink($target, &$text) {
1908                 # Valid link forms:
1909                 # Foobar -- normal
1910                 # :Foobar -- override special treatment of prefix (images, language links)
1911                 # /Foobar -- convert to CurrentPage/Foobar
1912                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1913                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1914                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1915
1916                 $fname = 'Parser::maybeDoSubpageLink';
1917                 wfProfileIn( $fname );
1918                 $ret = $target; # default return value is no change
1919
1920                 # bug 7425
1921                 $target = trim( $target );
1922
1923                 # Some namespaces don't allow subpages,
1924                 # so only perform processing if subpages are allowed
1925                 if( $this->areSubpagesAllowed() ) {
1926                         # Look at the first character
1927                         if( $target != '' && $target{0} == '/' ) {
1928                                 # / at end means we don't want the slash to be shown
1929                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1930                                 if( $trailingSlashes ) {
1931                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1932                                 } else {
1933                                         $noslash = substr( $target, 1 );
1934                                 }
1935
1936                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1937                                 if( '' === $text ) {
1938                                         $text = $target;
1939                                 } # this might be changed for ugliness reasons
1940                         } else {
1941                                 # check for .. subpage backlinks
1942                                 $dotdotcount = 0;
1943                                 $nodotdot = $target;
1944                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1945                                         ++$dotdotcount;
1946                                         $nodotdot = substr( $nodotdot, 3 );
1947                                 }
1948                                 if($dotdotcount > 0) {
1949                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1950                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1951                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1952                                                 # / at the end means don't show full path
1953                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1954                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1955                                                         if( '' === $text ) {
1956                                                                 $text = $nodotdot;
1957                                                         }
1958                                                 }
1959                                                 $nodotdot = trim( $nodotdot );
1960                                                 if( $nodotdot != '' ) {
1961                                                         $ret .= '/' . $nodotdot;
1962                                                 }
1963                                         }
1964                                 }
1965                         }
1966                 }
1967
1968                 wfProfileOut( $fname );
1969                 return $ret;
1970         }
1971
1972         /**#@+
1973          * Used by doBlockLevels()
1974          * @private
1975          */
1976         /* private */ function closeParagraph() {
1977                 $result = '';
1978                 if ( '' != $this->mLastSection ) {
1979                         $result = '</' . $this->mLastSection  . ">\n";
1980                 }
1981                 $this->mInPre = false;
1982                 $this->mLastSection = '';
1983                 return $result;
1984         }
1985         # getCommon() returns the length of the longest common substring
1986         # of both arguments, starting at the beginning of both.
1987         #
1988         /* private */ function getCommon( $st1, $st2 ) {
1989                 $fl = strlen( $st1 );
1990                 $shorter = strlen( $st2 );
1991                 if ( $fl < $shorter ) { $shorter = $fl; }
1992
1993                 for ( $i = 0; $i < $shorter; ++$i ) {
1994                         if ( $st1{$i} != $st2{$i} ) { break; }
1995                 }
1996                 return $i;
1997         }
1998         # These next three functions open, continue, and close the list
1999         # element appropriate to the prefix character passed into them.
2000         #
2001         /* private */ function openList( $char ) {
2002                 $result = $this->closeParagraph();
2003
2004                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2005                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2006                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2007                 else if ( ';' == $char ) {
2008                         $result .= '<dl><dt>';
2009                         $this->mDTopen = true;
2010                 }
2011                 else { $result = '<!-- ERR 1 -->'; }
2012
2013                 return $result;
2014         }
2015
2016         /* private */ function nextItem( $char ) {
2017                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2018                 else if ( ':' == $char || ';' == $char ) {
2019                         $close = '</dd>';
2020                         if ( $this->mDTopen ) { $close = '</dt>'; }
2021                         if ( ';' == $char ) {
2022                                 $this->mDTopen = true;
2023                                 return $close . '<dt>';
2024                         } else {
2025                                 $this->mDTopen = false;
2026                                 return $close . '<dd>';
2027                         }
2028                 }
2029                 return '<!-- ERR 2 -->';
2030         }
2031
2032         /* private */ function closeList( $char ) {
2033                 if ( '*' == $char ) { $text = '</li></ul>'; }
2034                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2035                 else if ( ':' == $char ) {
2036                         if ( $this->mDTopen ) {
2037                                 $this->mDTopen = false;
2038                                 $text = '</dt></dl>';
2039                         } else {
2040                                 $text = '</dd></dl>';
2041                         }
2042                 }
2043                 else {  return '<!-- ERR 3 -->'; }
2044                 return $text."\n";
2045         }
2046         /**#@-*/
2047
2048         /**
2049          * Make lists from lines starting with ':', '*', '#', etc.
2050          *
2051          * @private
2052          * @return string the lists rendered as HTML
2053          */
2054         function doBlockLevels( $text, $linestart ) {
2055                 $fname = 'Parser::doBlockLevels';
2056                 wfProfileIn( $fname );
2057
2058                 # Parsing through the text line by line.  The main thing
2059                 # happening here is handling of block-level elements p, pre,
2060                 # and making lists from lines starting with * # : etc.
2061                 #
2062                 $textLines = explode( "\n", $text );
2063
2064                 $lastPrefix = $output = '';
2065                 $this->mDTopen = $inBlockElem = false;
2066                 $prefixLength = 0;
2067                 $paragraphStack = false;
2068
2069                 if ( !$linestart ) {
2070                         $output .= array_shift( $textLines );
2071                 }
2072                 foreach ( $textLines as $oLine ) {
2073                         $lastPrefixLength = strlen( $lastPrefix );
2074                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2075                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2076                         if ( !$this->mInPre ) {
2077                                 # Multiple prefixes may abut each other for nested lists.
2078                                 $prefixLength = strspn( $oLine, '*#:;' );
2079                                 $pref = substr( $oLine, 0, $prefixLength );
2080
2081                                 # eh?
2082                                 $pref2 = str_replace( ';', ':', $pref );
2083                                 $t = substr( $oLine, $prefixLength );
2084                                 $this->mInPre = !empty($preOpenMatch);
2085                         } else {
2086                                 # Don't interpret any other prefixes in preformatted text
2087                                 $prefixLength = 0;
2088                                 $pref = $pref2 = '';
2089                                 $t = $oLine;
2090                         }
2091
2092                         # List generation
2093                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2094                                 # Same as the last item, so no need to deal with nesting or opening stuff
2095                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2096                                 $paragraphStack = false;
2097
2098                                 if ( substr( $pref, -1 ) == ';') {
2099                                         # The one nasty exception: definition lists work like this:
2100                                         # ; title : definition text
2101                                         # So we check for : in the remainder text to split up the
2102                                         # title and definition, without b0rking links.
2103                                         $term = $t2 = '';
2104                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2105                                                 $t = $t2;
2106                                                 $output .= $term . $this->nextItem( ':' );
2107                                         }
2108                                 }
2109                         } elseif( $prefixLength || $lastPrefixLength ) {
2110                                 # Either open or close a level...
2111                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2112                                 $paragraphStack = false;
2113
2114                                 while( $commonPrefixLength < $lastPrefixLength ) {
2115                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2116                                         --$lastPrefixLength;
2117                                 }
2118                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2119                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2120                                 }
2121                                 while ( $prefixLength > $commonPrefixLength ) {
2122                                         $char = substr( $pref, $commonPrefixLength, 1 );
2123                                         $output .= $this->openList( $char );
2124
2125                                         if ( ';' == $char ) {
2126                                                 # FIXME: This is dupe of code above
2127                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2128                                                         $t = $t2;
2129                                                         $output .= $term . $this->nextItem( ':' );
2130                                                 }
2131                                         }
2132                                         ++$commonPrefixLength;
2133                                 }
2134                                 $lastPrefix = $pref2;
2135                         }
2136                         if( 0 == $prefixLength ) {
2137                                 wfProfileIn( "$fname-paragraph" );
2138                                 # No prefix (not in list)--go to paragraph mode
2139                                 // XXX: use a stack for nestable elements like span, table and div
2140                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2141                                 $closematch = preg_match(
2142                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2143                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2144                                 if ( $openmatch or $closematch ) {
2145                                         $paragraphStack = false;
2146                                         # TODO bug 5718: paragraph closed
2147                                         $output .= $this->closeParagraph();
2148                                         if ( $preOpenMatch and !$preCloseMatch ) {
2149                                                 $this->mInPre = true;
2150                                         }
2151                                         if ( $closematch ) {
2152                                                 $inBlockElem = false;
2153                                         } else {
2154                                                 $inBlockElem = true;
2155                                         }
2156                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2157                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2158                                                 // pre
2159                                                 if ($this->mLastSection != 'pre') {
2160                                                         $paragraphStack = false;
2161                                                         $output .= $this->closeParagraph().'<pre>';
2162                                                         $this->mLastSection = 'pre';
2163                                                 }
2164                                                 $t = substr( $t, 1 );
2165                                         } else {
2166                                                 // paragraph
2167                                                 if ( '' == trim($t) ) {
2168                                                         if ( $paragraphStack ) {
2169                                                                 $output .= $paragraphStack.'<br />';
2170                                                                 $paragraphStack = false;
2171                                                                 $this->mLastSection = 'p';
2172                                                         } else {
2173                                                                 if ($this->mLastSection != 'p' ) {
2174                                                                         $output .= $this->closeParagraph();
2175                                                                         $this->mLastSection = '';
2176                                                                         $paragraphStack = '<p>';
2177                                                                 } else {
2178                                                                         $paragraphStack = '</p><p>';
2179                                                                 }
2180                                                         }
2181                                                 } else {
2182                                                         if ( $paragraphStack ) {
2183                                                                 $output .= $paragraphStack;
2184                                                                 $paragraphStack = false;
2185                                                                 $this->mLastSection = 'p';
2186                                                         } else if ($this->mLastSection != 'p') {
2187                                                                 $output .= $this->closeParagraph().'<p>';
2188                                                                 $this->mLastSection = 'p';
2189                                                         }
2190                                                 }
2191                                         }
2192                                 }
2193                                 wfProfileOut( "$fname-paragraph" );
2194                         }
2195                         // somewhere above we forget to get out of pre block (bug 785)
2196                         if($preCloseMatch && $this->mInPre) {
2197                                 $this->mInPre = false;
2198                         }
2199                         if ($paragraphStack === false) {
2200                                 $output .= $t."\n";
2201                         }
2202                 }
2203                 while ( $prefixLength ) {
2204                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2205                         --$prefixLength;
2206                 }
2207                 if ( '' != $this->mLastSection ) {
2208                         $output .= '</' . $this->mLastSection . '>';
2209                         $this->mLastSection = '';
2210                 }
2211
2212                 wfProfileOut( $fname );
2213                 return $output;
2214         }
2215
2216         /**
2217          * Split up a string on ':', ignoring any occurences inside tags
2218          * to prevent illegal overlapping.
2219          * @param string $str the string to split
2220          * @param string &$before set to everything before the ':'
2221          * @param string &$after set to everything after the ':'
2222          * return string the position of the ':', or false if none found
2223          */
2224         function findColonNoLinks($str, &$before, &$after) {
2225                 $fname = 'Parser::findColonNoLinks';
2226                 wfProfileIn( $fname );
2227
2228                 $pos = strpos( $str, ':' );
2229                 if( $pos === false ) {
2230                         // Nothing to find!
2231                         wfProfileOut( $fname );
2232                         return false;
2233                 }
2234
2235                 $lt = strpos( $str, '<' );
2236                 if( $lt === false || $lt > $pos ) {
2237                         // Easy; no tag nesting to worry about
2238                         $before = substr( $str, 0, $pos );
2239                         $after = substr( $str, $pos+1 );
2240                         wfProfileOut( $fname );
2241                         return $pos;
2242                 }
2243
2244                 // Ugly state machine to walk through avoiding tags.
2245                 $state = MW_COLON_STATE_TEXT;
2246                 $stack = 0;
2247                 $len = strlen( $str );
2248                 for( $i = 0; $i < $len; $i++ ) {
2249                         $c = $str{$i};
2250
2251                         switch( $state ) {
2252                         // (Using the number is a performance hack for common cases)
2253                         case 0: // MW_COLON_STATE_TEXT:
2254                                 switch( $c ) {
2255                                 case "<":
2256                                         // Could be either a <start> tag or an </end> tag
2257                                         $state = MW_COLON_STATE_TAGSTART;
2258                                         break;
2259                                 case ":":
2260                                         if( $stack == 0 ) {
2261                                                 // We found it!
2262                                                 $before = substr( $str, 0, $i );
2263                                                 $after = substr( $str, $i + 1 );
2264                                                 wfProfileOut( $fname );
2265                                                 return $i;
2266                                         }
2267                                         // Embedded in a tag; don't break it.
2268                                         break;
2269                                 default:
2270                                         // Skip ahead looking for something interesting
2271                                         $colon = strpos( $str, ':', $i );
2272                                         if( $colon === false ) {
2273                                                 // Nothing else interesting
2274                                                 wfProfileOut( $fname );
2275                                                 return false;
2276                                         }
2277                                         $lt = strpos( $str, '<', $i );
2278                                         if( $stack === 0 ) {
2279                                                 if( $lt === false || $colon < $lt ) {
2280                                                         // We found it!
2281                                                         $before = substr( $str, 0, $colon );
2282                                                         $after = substr( $str, $colon + 1 );
2283                                                         wfProfileOut( $fname );
2284                                                         return $i;
2285                                                 }
2286                                         }
2287                                         if( $lt === false ) {
2288                                                 // Nothing else interesting to find; abort!
2289                                                 // We're nested, but there's no close tags left. Abort!
2290                                                 break 2;
2291                                         }
2292                                         // Skip ahead to next tag start
2293                                         $i = $lt;
2294                                         $state = MW_COLON_STATE_TAGSTART;
2295                                 }
2296                                 break;
2297                         case 1: // MW_COLON_STATE_TAG:
2298                                 // In a <tag>
2299                                 switch( $c ) {
2300                                 case ">":
2301                                         $stack++;
2302                                         $state = MW_COLON_STATE_TEXT;
2303                                         break;
2304                                 case "/":
2305                                         // Slash may be followed by >?
2306                                         $state = MW_COLON_STATE_TAGSLASH;
2307                                         break;
2308                                 default:
2309                                         // ignore
2310                                 }
2311                                 break;
2312                         case 2: // MW_COLON_STATE_TAGSTART:
2313                                 switch( $c ) {
2314                                 case "/":
2315                                         $state = MW_COLON_STATE_CLOSETAG;
2316                                         break;
2317                                 case "!":
2318                                         $state = MW_COLON_STATE_COMMENT;
2319                                         break;
2320                                 case ">":
2321                                         // Illegal early close? This shouldn't happen D:
2322                                         $state = MW_COLON_STATE_TEXT;
2323                                         break;
2324                                 default:
2325                                         $state = MW_COLON_STATE_TAG;
2326                                 }
2327                                 break;
2328                         case 3: // MW_COLON_STATE_CLOSETAG:
2329                                 // In a </tag>
2330                                 if( $c == ">" ) {
2331                                         $stack--;
2332                                         if( $stack < 0 ) {
2333                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2334                                                 wfProfileOut( $fname );
2335                                                 return false;
2336                                         }
2337                                         $state = MW_COLON_STATE_TEXT;
2338                                 }
2339                                 break;
2340                         case MW_COLON_STATE_TAGSLASH:
2341                                 if( $c == ">" ) {
2342                                         // Yes, a self-closed tag <blah/>
2343                                         $state = MW_COLON_STATE_TEXT;
2344                                 } else {
2345                                         // Probably we're jumping the gun, and this is an attribute
2346                                         $state = MW_COLON_STATE_TAG;
2347                                 }
2348                                 break;
2349                         case 5: // MW_COLON_STATE_COMMENT:
2350                                 if( $c == "-" ) {
2351                                         $state = MW_COLON_STATE_COMMENTDASH;
2352                                 }
2353                                 break;
2354                         case MW_COLON_STATE_COMMENTDASH:
2355                                 if( $c == "-" ) {
2356                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2357                                 } else {
2358                                         $state = MW_COLON_STATE_COMMENT;
2359                                 }
2360                                 break;
2361                         case MW_COLON_STATE_COMMENTDASHDASH:
2362                                 if( $c == ">" ) {
2363                                         $state = MW_COLON_STATE_TEXT;
2364                                 } else {
2365                                         $state = MW_COLON_STATE_COMMENT;
2366                                 }
2367                                 break;
2368                         default:
2369                                 throw new MWException( "State machine error in $fname" );
2370                         }
2371                 }
2372                 if( $stack > 0 ) {
2373                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2374                         return false;
2375                 }
2376                 wfProfileOut( $fname );
2377                 return false;
2378         }
2379
2380         /**
2381          * Return value of a magic variable (like PAGENAME)
2382          *
2383          * @private
2384          */
2385         function getVariableValue( $index ) {
2386                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2387
2388                 /**
2389                  * Some of these require message or data lookups and can be
2390                  * expensive to check many times.
2391                  */
2392                 static $varCache = array();
2393                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2394                         if ( isset( $varCache[$index] ) ) {
2395                                 return $varCache[$index];
2396                         }
2397                 }
2398
2399                 $ts = time();
2400                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2401
2402                 # Use the time zone
2403                 global $wgLocaltimezone;
2404                 if ( isset( $wgLocaltimezone ) ) {
2405                         $oldtz = getenv( 'TZ' );
2406                         putenv( 'TZ='.$wgLocaltimezone );
2407                 }
2408                 $localTimestamp = date( 'YmdHis', $ts );
2409                 $localMonth = date( 'm', $ts );
2410                 $localMonthName = date( 'n', $ts );
2411                 $localDay = date( 'j', $ts );
2412                 $localDay2 = date( 'd', $ts );
2413                 $localDayOfWeek = date( 'w', $ts );
2414                 $localWeek = date( 'W', $ts );
2415                 $localYear = date( 'Y', $ts );
2416                 $localHour = date( 'H', $ts );
2417                 if ( isset( $wgLocaltimezone ) ) {
2418                         putenv( 'TZ='.$oldtz );
2419                 }
2420
2421                 switch ( $index ) {
2422                         case 'currentmonth':
2423                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2424                         case 'currentmonthname':
2425                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2426                         case 'currentmonthnamegen':
2427                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2428                         case 'currentmonthabbrev':
2429                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2430                         case 'currentday':
2431                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2432                         case 'currentday2':
2433                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2434                         case 'localmonth':
2435                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2436                         case 'localmonthname':
2437                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2438                         case 'localmonthnamegen':
2439                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2440                         case 'localmonthabbrev':
2441                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2442                         case 'localday':
2443                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2444                         case 'localday2':
2445                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2446                         case 'pagename':
2447                                 return $this->mTitle->getText();
2448                         case 'pagenamee':
2449                                 return $this->mTitle->getPartialURL();
2450                         case 'fullpagename':
2451                                 return $this->mTitle->getPrefixedText();
2452                         case 'fullpagenamee':
2453                                 return $this->mTitle->getPrefixedURL();
2454                         case 'subpagename':
2455                                 return $this->mTitle->getSubpageText();
2456                         case 'subpagenamee':
2457                                 return $this->mTitle->getSubpageUrlForm();
2458                         case 'basepagename':
2459                                 return $this->mTitle->getBaseText();
2460                         case 'basepagenamee':
2461                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2462                         case 'talkpagename':
2463                                 if( $this->mTitle->canTalk() ) {
2464                                         $talkPage = $this->mTitle->getTalkPage();
2465                                         return $talkPage->getPrefixedText();
2466                                 } else {
2467                                         return '';
2468                                 }
2469                         case 'talkpagenamee':
2470                                 if( $this->mTitle->canTalk() ) {
2471                                         $talkPage = $this->mTitle->getTalkPage();
2472                                         return $talkPage->getPrefixedUrl();
2473                                 } else {
2474                                         return '';
2475                                 }
2476                         case 'subjectpagename':
2477                                 $subjPage = $this->mTitle->getSubjectPage();
2478                                 return $subjPage->getPrefixedText();
2479                         case 'subjectpagenamee':
2480                                 $subjPage = $this->mTitle->getSubjectPage();
2481                                 return $subjPage->getPrefixedUrl();
2482                         case 'revisionid':
2483                                 return $this->mRevisionId;
2484                         case 'revisionday':
2485                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2486                         case 'revisionday2':
2487                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2488                         case 'revisionmonth':
2489                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2490                         case 'revisionyear':
2491                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2492                         case 'revisiontimestamp':
2493                                 return $this->getRevisionTimestamp();
2494                         case 'namespace':
2495                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2496                         case 'namespacee':
2497                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2498                         case 'talkspace':
2499                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2500                         case 'talkspacee':
2501                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2502                         case 'subjectspace':
2503                                 return $this->mTitle->getSubjectNsText();
2504                         case 'subjectspacee':
2505                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2506                         case 'currentdayname':
2507                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2508                         case 'currentyear':
2509                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2510                         case 'currenttime':
2511                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2512                         case 'currenthour':
2513                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2514                         case 'currentweek':
2515                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2516                                 // int to remove the padding
2517                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2518                         case 'currentdow':
2519                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2520                         case 'localdayname':
2521                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2522                         case 'localyear':
2523                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2524                         case 'localtime':
2525                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2526                         case 'localhour':
2527                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2528                         case 'localweek':
2529                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2530                                 // int to remove the padding
2531                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2532                         case 'localdow':
2533                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2534                         case 'numberofarticles':
2535                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2536                         case 'numberoffiles':
2537                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2538                         case 'numberofusers':
2539                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2540                         case 'numberofpages':
2541                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2542                         case 'numberofadmins':
2543                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2544                         case 'numberofedits':
2545                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2546                         case 'currenttimestamp':
2547                                 return $varCache[$index] = wfTimestampNow();
2548                         case 'localtimestamp':
2549                                 return $varCache[$index] = $localTimestamp;
2550                         case 'currentversion':
2551                                 return $varCache[$index] = SpecialVersion::getVersion();
2552                         case 'sitename':
2553                                 return $wgSitename;
2554                         case 'server':
2555                                 return $wgServer;
2556                         case 'servername':
2557                                 return $wgServerName;
2558                         case 'scriptpath':
2559                                 return $wgScriptPath;
2560                         case 'directionmark':
2561                                 return $wgContLang->getDirMark();
2562                         case 'contentlanguage':
2563                                 global $wgContLanguageCode;
2564                                 return $wgContLanguageCode;
2565                         default:
2566                                 $ret = null;
2567                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2568                                         return $ret;
2569                                 else
2570                                         return null;
2571                 }
2572         }
2573
2574         /**
2575          * initialise the magic variables (like CURRENTMONTHNAME)
2576          *
2577          * @private
2578          */
2579         function initialiseVariables() {
2580                 $fname = 'Parser::initialiseVariables';
2581                 wfProfileIn( $fname );
2582                 $variableIDs = MagicWord::getVariableIDs();
2583
2584                 $this->mVariables = array();
2585                 foreach ( $variableIDs as $id ) {
2586                         $mw =& MagicWord::get( $id );
2587                         $mw->addToArray( $this->mVariables, $id );
2588                 }
2589                 wfProfileOut( $fname );
2590         }
2591
2592         /**
2593          * parse any parentheses in format ((title|part|part))
2594          * and call callbacks to get a replacement text for any found piece
2595          *
2596          * @param string $text The text to parse
2597          * @param array $callbacks rules in form:
2598          *     '{' => array(                            # opening parentheses
2599          *                                      'end' => '}',   # closing parentheses
2600          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2601          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2602          *                                                                )
2603          *                                      )
2604          *                                      'min' => 2,     # Minimum parenthesis count in cb
2605          *                                      'max' => 3,     # Maximum parenthesis count in cb
2606          * @private
2607          */
2608         function replace_callback ($text, $callbacks) {
2609                 wfProfileIn( __METHOD__ );
2610                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2611                 $lastOpeningBrace = -1;                 # last not closed parentheses
2612
2613                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2614
2615                 $i = 0;
2616                 while ( $i < strlen( $text ) ) {
2617                         # Find next opening brace, closing brace or pipe
2618                         if ( $lastOpeningBrace == -1 ) {
2619                                 $currentClosing = '';
2620                                 $search = $validOpeningBraces;
2621                         } else {
2622                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2623                                 $search = $validOpeningBraces . '|' . $currentClosing;
2624                         }
2625                         $rule = null;
2626                         $i += strcspn( $text, $search, $i );
2627                         if ( $i < strlen( $text ) ) {
2628                                 if ( $text[$i] == '|' ) {
2629                                         $found = 'pipe';
2630                                 } elseif ( $text[$i] == $currentClosing ) {
2631                                         $found = 'close';
2632                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2633                                         $found = 'open';
2634                                         $rule = $callbacks[$text[$i]];
2635                                 } else {
2636                                         # Some versions of PHP have a strcspn which stops on null characters
2637                                         # Ignore and continue
2638                                         ++$i;
2639                                         continue;
2640                                 }
2641                         } else {
2642                                 # All done
2643                                 break;
2644                         }
2645
2646                         if ( $found == 'open' ) {
2647                                 # found opening brace, let's add it to parentheses stack
2648                                 $piece = array('brace' => $text[$i],
2649                                                            'braceEnd' => $rule['end'],
2650                                                            'title' => '',
2651                                                            'parts' => null);
2652
2653                                 # count opening brace characters
2654                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2655                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2656                                 $i += $piece['count'];
2657
2658                                 # we need to add to stack only if opening brace count is enough for one of the rules
2659                                 if ( $piece['count'] >= $rule['min'] ) {
2660                                         $lastOpeningBrace ++;
2661                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2662                                 }
2663                         } elseif ( $found == 'close' ) {
2664                                 # lets check if it is enough characters for closing brace
2665                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2666                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2667
2668                                 # check for maximum matching characters (if there are 5 closing
2669                                 # characters, we will probably need only 3 - depending on the rules)
2670                                 $matchingCount = 0;
2671                                 $matchingCallback = null;
2672                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2673                                 if ( $count > $cbType['max'] ) {
2674                                         # The specified maximum exists in the callback array, unless the caller
2675                                         # has made an error
2676                                         $matchingCount = $cbType['max'];
2677                                 } else {
2678                                         # Count is less than the maximum
2679                                         # Skip any gaps in the callback array to find the true largest match
2680                                         # Need to use array_key_exists not isset because the callback can be null
2681                                         $matchingCount = $count;
2682                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2683                                                 --$matchingCount;
2684                                         }
2685                                 }
2686
2687                                 if ($matchingCount <= 0) {
2688                                         $i += $count;
2689                                         continue;
2690                                 }
2691                                 $matchingCallback = $cbType['cb'][$matchingCount];
2692
2693                                 # let's set a title or last part (if '|' was found)
2694                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2695                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2696                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2697                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2698                                 } else {
2699                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2700                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2701                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2702                                 }
2703
2704                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2705                                 $pieceEnd = $i + $matchingCount;
2706
2707                                 if( is_callable( $matchingCallback ) ) {
2708                                         $cbArgs = array (
2709                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2710                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2711                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2712                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2713                                                                          );
2714                                         # finally we can call a user callback and replace piece of text
2715                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2716                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2717                                         $i = $pieceStart + strlen($replaceWith);
2718                                 } else {
2719                                         # null value for callback means that parentheses should be parsed, but not replaced
2720                                         $i += $matchingCount;
2721                                 }
2722
2723                                 # reset last opening parentheses, but keep it in case there are unused characters
2724                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2725                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2726                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2727                                                            'title' => '',
2728                                                            'parts' => null,
2729                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2730                                 $openingBraceStack[$lastOpeningBrace--] = null;
2731
2732                                 if ($matchingCount < $piece['count']) {
2733                                         $piece['count'] -= $matchingCount;
2734                                         $piece['startAt'] -= $matchingCount;
2735                                         $piece['partStart'] = $piece['startAt'];
2736                                         # do we still qualify for any callback with remaining count?
2737                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2738                                         while ( $piece['count'] ) {
2739                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2740                                                         $lastOpeningBrace++;
2741                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2742                                                         break;
2743                                                 }
2744                                                 --$piece['count'];
2745                                         }
2746                                 }
2747                         } elseif ( $found == 'pipe' ) {
2748                                 # lets set a title if it is a first separator, or next part otherwise
2749                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2750                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2751                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2752                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2753                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2754                                 } else {
2755                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2756                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2757                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2758                                 }
2759                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2760                         }
2761                 }
2762
2763                 wfProfileOut( __METHOD__ );
2764                 return $text;
2765         }
2766
2767         /**
2768          * Replace magic variables, templates, and template arguments
2769          * with the appropriate text. Templates are substituted recursively,
2770          * taking care to avoid infinite loops.
2771          *
2772          * Note that the substitution depends on value of $mOutputType:
2773          *  OT_WIKI: only {{subst:}} templates
2774          *  OT_MSG: only magic variables
2775          *  OT_HTML: all templates and magic variables
2776          *
2777          * @param string $tex The text to transform
2778          * @param array $args Key-value pairs representing template parameters to substitute
2779          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2780          * @private
2781          */
2782         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2783                 # Prevent too big inclusions
2784                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2785                         return $text;
2786                 }
2787
2788                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2789                 wfProfileIn( $fname );
2790
2791                 # This function is called recursively. To keep track of arguments we need a stack:
2792                 array_push( $this->mArgStack, $args );
2793
2794                 $braceCallbacks = array();
2795                 if ( !$argsOnly ) {
2796                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2797                 }
2798                 if ( $this->mOutputType != OT_MSG ) {
2799                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2800                 }
2801                 if ( $braceCallbacks ) {
2802                         $callbacks = array(
2803                                 '{' => array(
2804                                         'end' => '}',
2805                                         'cb' => $braceCallbacks,
2806                                         'min' => $argsOnly ? 3 : 2,
2807                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2808                                 ),
2809                                 '[' => array(
2810                                         'end' => ']',
2811                                         'cb' => array(2=>null),
2812                                         'min' => 2,
2813                                         'max' => 2,
2814                                 )
2815                         );
2816                         $text = $this->replace_callback ($text, $callbacks);
2817
2818                         array_pop( $this->mArgStack );
2819                 }
2820                 wfProfileOut( $fname );
2821                 return $text;
2822         }
2823
2824         /**
2825          * Replace magic variables
2826          * @private
2827          */
2828         function variableSubstitution( $matches ) {
2829                 global $wgContLang;
2830                 $fname = 'Parser::variableSubstitution';
2831                 $varname = $wgContLang->lc($matches[1]);
2832                 wfProfileIn( $fname );
2833                 $skip = false;
2834                 if ( $this->mOutputType == OT_WIKI ) {
2835                         # Do only magic variables prefixed by SUBST
2836                         $mwSubst =& MagicWord::get( 'subst' );
2837                         if (!$mwSubst->matchStartAndRemove( $varname ))
2838                                 $skip = true;
2839                         # Note that if we don't substitute the variable below,
2840                         # we don't remove the {{subst:}} magic word, in case
2841                         # it is a template rather than a magic variable.
2842                 }
2843                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2844                         $id = $this->mVariables[$varname];
2845                         # Now check if we did really match, case sensitive or not
2846                         $mw =& MagicWord::get( $id );
2847                         if ($mw->match($matches[1])) {
2848                                 $text = $this->getVariableValue( $id );
2849                                 $this->mOutput->mContainsOldMagic = true;
2850                         } else {
2851                                 $text = $matches[0];
2852                         }
2853                 } else {
2854                         $text = $matches[0];
2855                 }
2856                 wfProfileOut( $fname );
2857                 return $text;
2858         }
2859
2860
2861         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2862         static function createAssocArgs( $args ) {
2863                 $assocArgs = array();
2864                 $index = 1;
2865                 foreach( $args as $arg ) {
2866                         $eqpos = strpos( $arg, '=' );
2867                         if ( $eqpos === false ) {
2868                                 $assocArgs[$index++] = $arg;
2869                         } else {
2870                                 $name = trim( substr( $arg, 0, $eqpos ) );
2871                                 $value = trim( substr( $arg, $eqpos+1 ) );
2872                                 if ( $value === false ) {
2873                                         $value = '';
2874                                 }
2875                                 if ( $name !== false ) {
2876                                         $assocArgs[$name] = $value;
2877                                 }
2878                         }
2879                 }
2880
2881                 return $assocArgs;
2882         }
2883
2884         /**
2885          * Return the text of a template, after recursively
2886          * replacing any variables or templates within the template.
2887          *
2888          * @param array $piece The parts of the template
2889          *  $piece['text']: matched text
2890          *  $piece['title']: the title, i.e. the part before the |
2891          *  $piece['parts']: the parameter array
2892          * @return string the text of the template
2893          * @private
2894          */
2895         function braceSubstitution( $piece ) {
2896                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2897                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2898                 wfProfileIn( $fname );
2899                 wfProfileIn( __METHOD__.'-setup' );
2900
2901                 # Flags
2902                 $found = false;             # $text has been filled
2903                 $nowiki = false;            # wiki markup in $text should be escaped
2904                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2905                 $noargs = false;            # Don't replace triple-brace arguments in $text
2906                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2907                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2908                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2909                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2910
2911                 # Title object, where $text came from
2912                 $title = NULL;
2913
2914                 $linestart = '';
2915
2916
2917                 # $part1 is the bit before the first |, and must contain only title characters
2918                 # $args is a list of arguments, starting from index 0, not including $part1
2919
2920                 $titleText = $part1 = $piece['title'];
2921                 # If the third subpattern matched anything, it will start with |
2922
2923                 if (null == $piece['parts']) {
2924                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2925                         if ($replaceWith != $piece['text']) {
2926                                 $text = $replaceWith;
2927                                 $found = true;
2928                                 $noparse = true;
2929                                 $noargs = true;
2930                         }
2931                 }
2932
2933                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2934                 wfProfileOut( __METHOD__.'-setup' );
2935
2936                 # SUBST
2937                 wfProfileIn( __METHOD__.'-modifiers' );
2938                 if ( !$found ) {
2939                         $mwSubst =& MagicWord::get( 'subst' );
2940                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2941                                 # One of two possibilities is true:
2942                                 # 1) Found SUBST but not in the PST phase
2943                                 # 2) Didn't find SUBST and in the PST phase
2944                                 # In either case, return without further processing
2945                                 $text = $piece['text'];
2946                                 $found = true;
2947                                 $noparse = true;
2948                                 $noargs = true;
2949                         }
2950                 }
2951
2952                 # MSG, MSGNW and RAW
2953                 if ( !$found ) {
2954                         # Check for MSGNW:
2955                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2956                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2957                                 $nowiki = true;
2958                         } else {
2959                                 # Remove obsolete MSG:
2960                                 $mwMsg =& MagicWord::get( 'msg' );
2961                                 $mwMsg->matchStartAndRemove( $part1 );
2962                         }
2963
2964                         # Check for RAW:
2965                         $mwRaw =& MagicWord::get( 'raw' );
2966                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2967                                 $forceRawInterwiki = true;
2968                         }
2969                 }
2970                 wfProfileOut( __METHOD__.'-modifiers' );
2971
2972                 //save path level before recursing into functions & templates.
2973                 $lastPathLevel = $this->mTemplatePath;
2974
2975                 # Parser functions
2976                 if ( !$found ) {
2977                         wfProfileIn( __METHOD__ . '-pfunc' );
2978
2979                         $colonPos = strpos( $part1, ':' );
2980                         if ( $colonPos !== false ) {
2981                                 # Case sensitive functions
2982                                 $function = substr( $part1, 0, $colonPos );
2983                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2984                                         $function = $this->mFunctionSynonyms[1][$function];
2985                                 } else {
2986                                         # Case insensitive functions
2987                                         $function = strtolower( $function );
2988                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2989                                                 $function = $this->mFunctionSynonyms[0][$function];
2990                                         } else {
2991                                                 $function = false;
2992                                         }
2993                                 }
2994                                 if ( $function ) {
2995                                         $funcArgs = array_map( 'trim', $args );
2996                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2997                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2998                                         $found = true;
2999
3000                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3001                                         //$noargs = true;
3002                                         //$noparse = true;
3003
3004                                         if ( is_array( $result ) ) {
3005                                                 if ( isset( $result[0] ) ) {
3006                                                         $text = $linestart . $result[0];
3007                                                         unset( $result[0] );
3008                                                 }
3009
3010                                                 // Extract flags into the local scope
3011                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3012                                                 extract( $result );
3013                                         } else {
3014                                                 $text = $linestart . $result;
3015                                         }
3016                                 }
3017                         }
3018                         wfProfileOut( __METHOD__ . '-pfunc' );
3019                 }
3020
3021                 # Template table test
3022
3023                 # Did we encounter this template already? If yes, it is in the cache
3024                 # and we need to check for loops.
3025                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3026                         $found = true;
3027
3028                         # Infinite loop test
3029                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3030                                 $noparse = true;
3031                                 $noargs = true;
3032                                 $found = true;
3033                                 $text = $linestart .
3034                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3035                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3036                         } else {
3037                                 # set $text to cached message.
3038                                 $text = $linestart . $this->mTemplates[$piece['title']];
3039                                 #treat title for cached page the same as others
3040                                 $ns = NS_TEMPLATE;
3041                                 $subpage = '';
3042                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3043                                 if ($subpage !== '') {
3044                                   $ns = $this->mTitle->getNamespace();
3045                                 }
3046                                 $title = Title::newFromText( $part1, $ns );
3047                                 //used by include size checking
3048                                 $titleText = $title->getPrefixedText();
3049                                 //used by edit section links
3050                                 $replaceHeadings = true;
3051
3052                         }
3053                 }
3054
3055                 # Load from database
3056                 if ( !$found ) {
3057                         wfProfileIn( __METHOD__ . '-loadtpl' );
3058                         $ns = NS_TEMPLATE;
3059                         # declaring $subpage directly in the function call
3060                         # does not work correctly with references and breaks
3061                         # {{/subpage}}-style inclusions
3062                         $subpage = '';
3063                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3064                         if ($subpage !== '') {
3065                                 $ns = $this->mTitle->getNamespace();
3066                         }
3067                         $title = Title::newFromText( $part1, $ns );
3068
3069
3070                         if ( !is_null( $title ) ) {
3071                                 $titleText = $title->getPrefixedText();
3072                                 # Check for language variants if the template is not found
3073                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3074                                         $wgContLang->findVariantLink($part1, $title);
3075                                 }
3076
3077                                 if ( !$title->isExternal() ) {
3078                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3079                                                 $text = SpecialPage::capturePath( $title );
3080                                                 if ( is_string( $text ) ) {
3081                                                         $found = true;
3082                                                         $noparse = true;
3083                                                         $noargs = true;
3084                                                         $isHTML = true;
3085                                                         $this->disableCache();
3086                                                 }
3087                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3088                                                 $found = false; //access denied
3089                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3090                                         } else {
3091                                                 $articleContent = $this->fetchTemplate( $title );
3092                                                 if ( $articleContent !== false ) {
3093                                                         $found = true;
3094                                                         $text = $articleContent;
3095                                                         $replaceHeadings = true;
3096                                                 }
3097                                         }
3098
3099                                         # If the title is valid but undisplayable, make a link to it
3100                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3101                                                 $text = "[[:$titleText]]";
3102                                                 $found = true;
3103                                         }
3104                                 } elseif ( $title->isTrans() ) {
3105                                         // Interwiki transclusion
3106                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3107                                                 $text = $this->interwikiTransclude( $title, 'render' );
3108                                                 $isHTML = true;
3109                                                 $noparse = true;
3110                                         } else {
3111                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3112                                                 $replaceHeadings = true;
3113                                         }
3114                                         $found = true;
3115                                 }
3116
3117                                 # Template cache array insertion
3118                                 # Use the original $piece['title'] not the mangled $part1, so that
3119                                 # modifiers such as RAW: produce separate cache entries
3120                                 if( $found ) {
3121                                         if( $isHTML ) {
3122                                                 // A special page; don't store it in the template cache.
3123                                         } else {
3124                                                 $this->mTemplates[$piece['title']] = $text;
3125                                         }
3126                                         $text = $linestart . $text;
3127                                 }
3128                         }
3129                         wfProfileOut( __METHOD__ . '-loadtpl' );
3130                 }
3131
3132                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3133                         # Error, oversize inclusion
3134                         $text = $linestart .
3135                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3136                         $noparse = true;
3137                         $noargs = true;
3138                 }
3139
3140                 # Recursive parsing, escaping and link table handling
3141                 # Only for HTML output
3142                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3143                         $text = wfEscapeWikiText( $text );
3144                 } elseif ( !$this->ot['msg'] && $found ) {
3145                         if ( $noargs ) {
3146                                 $assocArgs = array();
3147                         } else {
3148                                 # Clean up argument array
3149                                 $assocArgs = self::createAssocArgs($args);
3150                                 # Add a new element to the templace recursion path
3151                                 $this->mTemplatePath[$part1] = 1;
3152                         }
3153
3154                         if ( !$noparse ) {
3155                                 # If there are any <onlyinclude> tags, only include them
3156                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3157                                         $replacer = new OnlyIncludeReplacer;
3158                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3159                                                 array( &$replacer, 'replace' ), $text );
3160                                         $text = $replacer->output;
3161                                 }
3162                                 # Remove <noinclude> sections and <includeonly> tags
3163                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3164                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3165
3166                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3167                                         # Strip <nowiki>, <pre>, etc.
3168                                         $text = $this->strip( $text, $this->mStripState );
3169                                         if ( $this->ot['html'] ) {
3170                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3171                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3172                                                 $text = Sanitizer::removeHTMLcomments( $text );
3173                                         }
3174                                 }
3175                                 $text = $this->replaceVariables( $text, $assocArgs );
3176
3177                                 # If the template begins with a table or block-level
3178                                 # element, it should be treated as beginning a new line.
3179                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3180                                         $text = "\n" . $text;
3181                                 }
3182                         } elseif ( !$noargs ) {
3183                                 # $noparse and !$noargs
3184                                 # Just replace the arguments, not any double-brace items
3185                                 # This is used for rendered interwiki transclusion
3186                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3187                         }
3188                 }
3189                 # Prune lower levels off the recursion check path
3190                 $this->mTemplatePath = $lastPathLevel;
3191
3192                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3193                         # Error, oversize inclusion
3194                         $text = $linestart .
3195                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3196                         $noparse = true;
3197                         $noargs = true;
3198                 }
3199
3200                 if ( !$found ) {
3201                         wfProfileOut( $fname );
3202                         return $piece['text'];
3203                 } else {
3204                         wfProfileIn( __METHOD__ . '-placeholders' );
3205                         if ( $isHTML ) {
3206                                 # Replace raw HTML by a placeholder
3207                                 # Add a blank line preceding, to prevent it from mucking up
3208                                 # immediately preceding headings
3209                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3210                         } else {
3211                                 # replace ==section headers==
3212                                 # XXX this needs to go away once we have a better parser.
3213                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3214                                         if( !is_null( $title ) )
3215                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3216                                         else
3217                                                 $encodedname = base64_encode("");
3218                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3219                                                 PREG_SPLIT_DELIM_CAPTURE);
3220                                         $text = '';
3221                                         $nsec = $headingOffset;
3222                                         for( $i = 0; $i < count($m); $i += 2 ) {
3223                                                 $text .= $m[$i];
3224                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3225                                                 $hl = $m[$i + 1];
3226                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3227                                                         $text .= $hl;
3228                                                         continue;
3229                                                 }
3230                                                 $m2 = array();
3231                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3232                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3233                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3234
3235                                                 $nsec++;
3236                                         }
3237                                 }
3238                         }
3239                         wfProfileOut( __METHOD__ . '-placeholders' );
3240                 }
3241
3242                 # Prune lower levels off the recursion check path
3243                 $this->mTemplatePath = $lastPathLevel;
3244
3245                 if ( !$found ) {
3246                         wfProfileOut( $fname );
3247                         return $piece['text'];
3248                 } else {
3249                         wfProfileOut( $fname );
3250                         return $text;
3251                 }
3252         }
3253
3254         /**
3255          * Fetch the unparsed text of a template and register a reference to it.
3256          */
3257         function fetchTemplate( $title ) {
3258                 $text = false;
3259                 // Loop to fetch the article, with up to 1 redirect
3260                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3261                         $rev = Revision::newFromTitle( $title );
3262                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3263                         if ( $rev ) {
3264                                 $text = $rev->getText();
3265                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3266                                 global $wgLang;
3267                                 $message = $wgLang->lcfirst( $title->getText() );
3268                                 $text = wfMsgForContentNoTrans( $message );
3269                                 if( wfEmptyMsg( $message, $text ) ) {
3270                                         $text = false;
3271                                         break;
3272                                 }
3273                         } else {
3274                                 break;
3275                         }
3276                         if ( $text === false ) {
3277                                 break;
3278                         }
3279                         // Redirect?
3280                         $title = Title::newFromRedirect( $text );
3281                 }
3282                 return $text;
3283         }
3284
3285         /**
3286          * Transclude an interwiki link.
3287          */
3288         function interwikiTransclude( $title, $action ) {
3289                 global $wgEnableScaryTranscluding;
3290
3291                 if (!$wgEnableScaryTranscluding)
3292                         return wfMsg('scarytranscludedisabled');
3293
3294                 $url = $title->getFullUrl( "action=$action" );
3295
3296                 if (strlen($url) > 255)
3297                         return wfMsg('scarytranscludetoolong');
3298                 return $this->fetchScaryTemplateMaybeFromCache($url);
3299         }
3300
3301         function fetchScaryTemplateMaybeFromCache($url) {
3302                 global $wgTranscludeCacheExpiry;
3303                 $dbr = wfGetDB(DB_SLAVE);
3304                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3305                                 array('tc_url' => $url));
3306                 if ($obj) {
3307                         $time = $obj->tc_time;
3308                         $text = $obj->tc_contents;
3309                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3310                                 return $text;
3311                         }
3312                 }
3313
3314                 $text = Http::get($url);
3315                 if (!$text)
3316                         return wfMsg('scarytranscludefailed', $url);
3317
3318                 $dbw = wfGetDB(DB_MASTER);
3319                 $dbw->replace('transcache', array('tc_url'), array(
3320                         'tc_url' => $url,
3321                         'tc_time' => time(),
3322                         'tc_contents' => $text));
3323                 return $text;
3324         }
3325
3326
3327         /**
3328          * Triple brace replacement -- used for template arguments
3329          * @private
3330          */
3331         function argSubstitution( $matches ) {
3332                 $arg = trim( $matches['title'] );
3333                 $text = $matches['text'];
3334                 $inputArgs = end( $this->mArgStack );
3335
3336                 if ( array_key_exists( $arg, $inputArgs ) ) {
3337                         $text = $inputArgs[$arg];
3338                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3339                 null != $matches['parts'] && count($matches['parts']) > 0) {
3340                         $text = $matches['parts'][0];
3341                 }
3342                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3343                         $text = $matches['text'] .
3344                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3345                 }
3346
3347                 return $text;
3348         }
3349
3350         /**
3351          * Increment an include size counter
3352          *
3353          * @param string $type The type of expansion
3354          * @param integer $size The size of the text
3355          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3356          */
3357         function incrementIncludeSize( $type, $size ) {
3358                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3359                         return false;
3360                 } else {
3361                         $this->mIncludeSizes[$type] += $size;
3362                         return true;
3363                 }
3364         }
3365
3366         /**
3367          * Detect __NOGALLERY__ magic word and set a placeholder
3368          */
3369         function stripNoGallery( &$text ) {
3370                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3371                 # do not add TOC
3372                 $mw = MagicWord::get( 'nogallery' );
3373                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3374         }
3375
3376         /**
3377          * Detect __TOC__ magic word and set a placeholder
3378          */
3379         function stripToc( $text ) {
3380                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3381                 # do not add TOC
3382                 $mw = MagicWord::get( 'notoc' );
3383                 if( $mw->matchAndRemove( $text ) ) {
3384                         $this->mShowToc = false;
3385                 }
3386
3387                 $mw = MagicWord::get( 'toc' );
3388                 if( $mw->match( $text ) ) {
3389                         $this->mShowToc = true;
3390                         $this->mForceTocPosition = true;
3391
3392                         // Set a placeholder. At the end we'll fill it in with the TOC.
3393                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3394
3395                         // Only keep the first one.
3396                         $text = $mw->replace( '', $text );
3397                 }
3398                 return $text;
3399         }
3400
3401         /**
3402          * This function accomplishes several tasks:
3403          * 1) Auto-number headings if that option is enabled
3404          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3405          * 3) Add a Table of contents on the top for users who have enabled the option
3406          * 4) Auto-anchor headings
3407          *
3408          * It loops through all headlines, collects the necessary data, then splits up the
3409          * string and re-inserts the newly formatted headlines.
3410          *
3411          * @param string $text
3412          * @param boolean $isMain
3413          * @private
3414          */
3415         function formatHeadings( $text, $isMain=true ) {
3416                 global $wgMaxTocLevel, $wgContLang;
3417
3418                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3419                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3420                         $showEditLink = 0;
3421                 } else {
3422                         $showEditLink = $this->mOptions->getEditSection();
3423                 }
3424
3425                 # Inhibit editsection links if requested in the page
3426                 $esw =& MagicWord::get( 'noeditsection' );
3427                 if( $esw->matchAndRemove( $text ) ) {
3428                         $showEditLink = 0;
3429                 }
3430
3431                 # Get all headlines for numbering them and adding funky stuff like [edit]
3432                 # links - this is for later, but we need the number of headlines right now
3433                 $matches = array();
3434                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3435
3436                 # if there are fewer than 4 headlines in the article, do not show TOC
3437                 # unless it's been explicitly enabled.
3438                 $enoughToc = $this->mShowToc &&
3439                         (($numMatches >= 4) || $this->mForceTocPosition);
3440
3441                 # Allow user to stipulate that a page should have a "new section"
3442                 # link added via __NEWSECTIONLINK__
3443                 $mw =& MagicWord::get( 'newsectionlink' );
3444                 if( $mw->matchAndRemove( $text ) )
3445                         $this->mOutput->setNewSection( true );
3446
3447                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3448                 # override above conditions and always show TOC above first header
3449                 $mw =& MagicWord::get( 'forcetoc' );
3450                 if ($mw->matchAndRemove( $text ) ) {
3451                         $this->mShowToc = true;
3452                         $enoughToc = true;
3453                 }
3454
3455                 # Never ever show TOC if no headers
3456                 if( $numMatches < 1 ) {
3457                         $enoughToc = false;
3458                 }
3459
3460                 # We need this to perform operations on the HTML
3461                 $sk = $this->mOptions->getSkin();
3462
3463                 # headline counter
3464                 $headlineCount = 0;
3465                 $sectionCount = 0; # headlineCount excluding template sections
3466
3467                 # Ugh .. the TOC should have neat indentation levels which can be
3468                 # passed to the skin functions. These are determined here
3469                 $toc = '';
3470                 $full = '';
3471                 $head = array();
3472                 $sublevelCount = array();
3473                 $levelCount = array();
3474                 $toclevel = 0;
3475                 $level = 0;
3476                 $prevlevel = 0;
3477                 $toclevel = 0;
3478                 $prevtoclevel = 0;
3479
3480                 foreach( $matches[3] as $headline ) {
3481                         $istemplate = 0;
3482                         $templatetitle = '';
3483                         $templatesection = 0;
3484                         $numbering = '';
3485                         $mat = array();
3486                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3487                                 $istemplate = 1;
3488                                 $templatetitle = base64_decode($mat[1]);
3489                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3490                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3491                         }
3492
3493                         if( $toclevel ) {
3494                                 $prevlevel = $level;
3495                                 $prevtoclevel = $toclevel;
3496                         }
3497                         $level = $matches[1][$headlineCount];
3498
3499                         if( $doNumberHeadings || $enoughToc ) {
3500
3501                                 if ( $level > $prevlevel ) {
3502                                         # Increase TOC level
3503                                         $toclevel++;
3504                                         $sublevelCount[$toclevel] = 0;
3505                                         if( $toclevel<$wgMaxTocLevel ) {
3506                                                 $toc .= $sk->tocIndent();
3507                                         }
3508                                 }
3509                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3510                                         # Decrease TOC level, find level to jump to
3511
3512                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3513                                                 # Can only go down to level 1
3514                                                 $toclevel = 1;
3515                                         } else {
3516                                                 for ($i = $toclevel; $i > 0; $i--) {
3517                                                         if ( $levelCount[$i] == $level ) {
3518                                                                 # Found last matching level
3519                                                                 $toclevel = $i;
3520                                                                 break;
3521                                                         }
3522                                                         elseif ( $levelCount[$i] < $level ) {
3523                                                                 # Found first matching level below current level
3524                                                                 $toclevel = $i + 1;
3525                                                                 break;
3526                                                         }
3527                                                 }
3528                                         }
3529                                         if( $toclevel<$wgMaxTocLevel ) {
3530                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3531                                         }
3532                                 }
3533                                 else {
3534                                         # No change in level, end TOC line
3535                                         if( $toclevel<$wgMaxTocLevel ) {
3536                                                 $toc .= $sk->tocLineEnd();
3537                                         }
3538                                 }
3539
3540                                 $levelCount[$toclevel] = $level;
3541
3542                                 # count number of headlines for each level
3543                                 @$sublevelCount[$toclevel]++;
3544                                 $dot = 0;
3545                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3546                                         if( !empty( $sublevelCount[$i] ) ) {
3547                                                 if( $dot ) {
3548                                                         $numbering .= '.';
3549                                                 }
3550                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3551                                                 $dot = 1;
3552                                         }
3553                                 }
3554                         }
3555
3556                         # The canonized header is a version of the header text safe to use for links
3557                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3558                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3559
3560                         # Remove link placeholders by the link text.
3561                         #     <!--LINK number-->
3562                         # turns into
3563                         #     link text with suffix
3564                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3565                                                             "\$this->mLinkHolders['texts'][\$1]",
3566                                                             $canonized_headline );
3567                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3568                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3569                                                             $canonized_headline );
3570
3571                         # strip out HTML
3572                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3573                         $tocline = trim( $canonized_headline );
3574                         # Save headline for section edit hint before it's escaped
3575                         $headline_hint = trim( $canonized_headline );
3576                         $canonized_headline = Sanitizer::escapeId( $tocline );
3577                         $refers[$headlineCount] = $canonized_headline;
3578
3579                         # count how many in assoc. array so we can track dupes in anchors
3580                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3581                         $refcount[$headlineCount]=$refers[$canonized_headline];
3582
3583                         # Don't number the heading if it is the only one (looks silly)
3584                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3585                                 # the two are different if the line contains a link
3586                                 $headline=$numbering . ' ' . $headline;
3587                         }
3588
3589                         # Create the anchor for linking from the TOC to the section
3590                         $anchor = $canonized_headline;
3591                         if($refcount[$headlineCount] > 1 ) {
3592                                 $anchor .= '_' . $refcount[$headlineCount];
3593                         }
3594                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3595                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3596                         }
3597                         # give headline the correct <h#> tag
3598                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3599                                 if( $istemplate )
3600                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3601                                 else
3602                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3603                         } else {
3604                                 $editlink = '';
3605                         }
3606                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3607
3608                         $headlineCount++;
3609                         if( !$istemplate )
3610                                 $sectionCount++;
3611                 }
3612
3613                 if( $enoughToc ) {
3614                         if( $toclevel<$wgMaxTocLevel ) {
3615                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3616                         }
3617                         $toc = $sk->tocList( $toc );
3618                 }
3619
3620                 # split up and insert constructed headlines
3621
3622                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3623                 $i = 0;
3624
3625                 foreach( $blocks as $block ) {
3626                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3627                                 # This is the [edit] link that appears for the top block of text when
3628                                 # section editing is enabled
3629
3630                                 # Disabled because it broke block formatting
3631                                 # For example, a bullet point in the top line
3632                                 # $full .= $sk->editSectionLink(0);
3633                         }
3634                         $full .= $block;
3635                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3636                                 # Top anchor now in skin
3637                                 $full = $full.$toc;
3638                         }
3639
3640                         if( !empty( $head[$i] ) ) {
3641                                 $full .= $head[$i];
3642                         }
3643                         $i++;
3644                 }
3645                 if( $this->mForceTocPosition ) {
3646                         return str_replace( '<!--MWTOC-->', $toc, $full );
3647                 } else {
3648                         return $full;
3649                 }
3650         }
3651
3652         /**
3653          * Transform wiki markup when saving a page by doing \r\n -> \n
3654          * conversion, substitting signatures, {{subst:}} templates, etc.
3655          *
3656          * @param string $text the text to transform
3657          * @param Title &$title the Title object for the current article
3658          * @param User &$user the User object describing the current user
3659          * @param ParserOptions $options parsing options
3660          * @param bool $clearState whether to clear the parser state first
3661          * @return string the altered wiki markup
3662          * @public
3663          */
3664         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3665                 $this->mOptions = $options;
3666                 $this->mTitle =& $title;
3667                 $this->setOutputType( OT_WIKI );
3668
3669                 if ( $clearState ) {
3670                         $this->clearState();
3671                 }
3672
3673                 $stripState = new StripState;
3674                 $pairs = array(
3675                         "\r\n" => "\n",
3676                 );
3677                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3678                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3679                 $text = $this->pstPass2( $text, $stripState, $user );
3680                 $text = $stripState->unstripBoth( $text );
3681                 return $text;
3682         }
3683
3684         /**
3685          * Pre-save transform helper function
3686          * @private
3687          */
3688         function pstPass2( $text, &$stripState, $user ) {
3689                 global $wgContLang, $wgLocaltimezone;
3690
3691                 /* Note: This is the timestamp saved as hardcoded wikitext to
3692                  * the database, we use $wgContLang here in order to give
3693                  * everyone the same signature and use the default one rather
3694                  * than the one selected in each user's preferences.
3695                  */
3696                 if ( isset( $wgLocaltimezone ) ) {
3697                         $oldtz = getenv( 'TZ' );
3698                         putenv( 'TZ='.$wgLocaltimezone );
3699                 }
3700                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3701                   ' (' . date( 'T' ) . ')';
3702                 if ( isset( $wgLocaltimezone ) ) {
3703                         putenv( 'TZ='.$oldtz );
3704                 }
3705
3706                 # Variable replacement
3707                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3708                 $text = $this->replaceVariables( $text );
3709
3710                 # Strip out <nowiki> etc. added via replaceVariables
3711                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3712
3713                 # Signatures
3714                 $sigText = $this->getUserSig( $user );
3715                 $text = strtr( $text, array(
3716                         '~~~~~' => $d,
3717                         '~~~~' => "$sigText $d",
3718                         '~~~' => $sigText
3719                 ) );
3720
3721                 # Context links: [[|name]] and [[name (context)|]]
3722                 #
3723                 global $wgLegalTitleChars;
3724                 $tc = "[$wgLegalTitleChars]";
3725                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3726
3727                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3728                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3729                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3730
3731                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3732                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3733                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3734
3735                 $t = $this->mTitle->getText();
3736                 $m = array();
3737                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3738                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3739                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3740                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3741                 } else {
3742                         # if there's no context, don't bother duplicating the title
3743                         $text = preg_replace( $p2, '[[\\1]]', $text );
3744                 }
3745
3746                 # Trim trailing whitespace
3747                 $text = rtrim( $text );
3748
3749                 return $text;
3750         }
3751
3752         /**
3753          * Fetch the user's signature text, if any, and normalize to
3754          * validated, ready-to-insert wikitext.
3755          *
3756          * @param User $user
3757          * @return string
3758          * @private
3759          */
3760         function getUserSig( &$user ) {
3761                 $username = $user->getName();
3762                 $nickname = $user->getOption( 'nickname' );
3763                 $nickname = $nickname === '' ? $username : $nickname;
3764
3765                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3766                         # Sig. might contain markup; validate this
3767                         if( $this->validateSig( $nickname ) !== false ) {
3768                                 # Validated; clean up (if needed) and return it
3769                                 return $this->cleanSig( $nickname, true );
3770                         } else {
3771                                 # Failed to validate; fall back to the default
3772                                 $nickname = $username;
3773                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3774                         }
3775                 }
3776
3777                 // Make sure nickname doesnt get a sig in a sig
3778                 $nickname = $this->cleanSigInSig( $nickname );
3779
3780                 # If we're still here, make it a link to the user page
3781                 $userpage = $user->getUserPage();
3782                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3783         }
3784
3785         /**
3786          * Check that the user's signature contains no bad XML
3787          *
3788          * @param string $text
3789          * @return mixed An expanded string, or false if invalid.
3790          */
3791         function validateSig( $text ) {
3792                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3793         }
3794
3795         /**
3796          * Clean up signature text
3797          *
3798          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3799          * 2) Substitute all transclusions
3800          *
3801          * @param string $text
3802          * @param $parsing Whether we're cleaning (preferences save) or parsing
3803          * @return string Signature text
3804          */
3805         function cleanSig( $text, $parsing = false ) {
3806                 global $wgTitle;
3807                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3808
3809                 $substWord = MagicWord::get( 'subst' );
3810                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3811                 $substText = '{{' . $substWord->getSynonym( 0 );
3812
3813                 $text = preg_replace( $substRegex, $substText, $text );
3814                 $text = $this->cleanSigInSig( $text );
3815                 $text = $this->replaceVariables( $text );
3816
3817                 $this->clearState();
3818                 return $text;
3819         }
3820
3821         /**
3822          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3823          * @param string $text
3824          * @return string Signature text with /~{3,5}/ removed
3825          */
3826         function cleanSigInSig( $text ) {
3827                 $text = preg_replace( '/~{3,5}/', '', $text );
3828                 return $text;
3829         }
3830
3831         /**
3832          * Set up some variables which are usually set up in parse()
3833          * so that an external function can call some class members with confidence
3834          * @public
3835          */
3836         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3837                 $this->mTitle =& $title;
3838                 $this->mOptions = $options;
3839                 $this->setOutputType( $outputType );
3840                 if ( $clearState ) {
3841                         $this->clearState();
3842                 }
3843         }
3844
3845         /**
3846          * Transform a MediaWiki message by replacing magic variables.
3847          *
3848          * @param string $text the text to transform
3849          * @param ParserOptions $options  options
3850          * @return string the text with variables substituted
3851          * @public
3852          */
3853         function transformMsg( $text, $options ) {
3854                 global $wgTitle;
3855                 static $executing = false;
3856
3857                 $fname = "Parser::transformMsg";
3858
3859                 # Guard against infinite recursion
3860                 if ( $executing ) {
3861                         return $text;
3862                 }
3863                 $executing = true;
3864
3865                 wfProfileIn($fname);
3866
3867                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3868                         $this->mTitle = $wgTitle;
3869                 } else {
3870                         $this->mTitle = Title::newFromText('msg');
3871                 }
3872                 $this->mOptions = $options;
3873                 $this->setOutputType( OT_MSG );
3874                 $this->clearState();
3875                 $text = $this->replaceVariables( $text );
3876
3877                 $executing = false;
3878                 wfProfileOut($fname);
3879                 return $text;
3880         }
3881
3882         /**
3883          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3884          * The callback should have the following form:
3885          *    function myParserHook( $text, $params, &$parser ) { ... }
3886          *
3887          * Transform and return $text. Use $parser for any required context, e.g. use
3888          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3889          *
3890          * @public
3891          *
3892          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3893          * @param mixed $callback The callback function (and object) to use for the tag
3894          *
3895          * @return The old value of the mTagHooks array associated with the hook
3896          */
3897         function setHook( $tag, $callback ) {
3898                 $tag = strtolower( $tag );
3899                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3900                 $this->mTagHooks[$tag] = $callback;
3901
3902                 return $oldVal;
3903         }
3904
3905         /**
3906          * Create a function, e.g. {{sum:1|2|3}}
3907          * The callback function should have the form:
3908          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3909          *
3910          * The callback may either return the text result of the function, or an array with the text
3911          * in element 0, and a number of flags in the other elements. The names of the flags are
3912          * specified in the keys. Valid flags are:
3913          *   found                     The text returned is valid, stop processing the template. This
3914          *                             is on by default.
3915          *   nowiki                    Wiki markup in the return value should be escaped
3916          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3917          *   noargs                    Don't replace triple-brace arguments in the return value
3918          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3919          *
3920          * @public
3921          *
3922          * @param string $id The magic word ID
3923          * @param mixed $callback The callback function (and object) to use
3924          * @param integer $flags a combination of the following flags:
3925          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3926          *
3927          * @return The old callback function for this name, if any
3928          */
3929         function setFunctionHook( $id, $callback, $flags = 0 ) {
3930                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3931                 $this->mFunctionHooks[$id] = $callback;
3932
3933                 # Add to function cache
3934                 $mw = MagicWord::get( $id );
3935                 if( !$mw )
3936                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3937
3938                 $synonyms = $mw->getSynonyms();
3939                 $sensitive = intval( $mw->isCaseSensitive() );
3940
3941                 foreach ( $synonyms as $syn ) {
3942                         # Case
3943                         if ( !$sensitive ) {
3944                                 $syn = strtolower( $syn );
3945                         }
3946                         # Add leading hash
3947                         if ( !( $flags & SFH_NO_HASH ) ) {
3948                                 $syn = '#' . $syn;
3949                         }
3950                         # Remove trailing colon
3951                         if ( substr( $syn, -1, 1 ) == ':' ) {
3952                                 $syn = substr( $syn, 0, -1 );
3953                         }
3954                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3955                 }
3956                 return $oldVal;
3957         }
3958
3959         /**
3960          * Get all registered function hook identifiers
3961          *
3962          * @return array
3963          */
3964         function getFunctionHooks() {
3965                 return array_keys( $this->mFunctionHooks );
3966         }
3967
3968         /**
3969          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3970          * Placeholders created in Skin::makeLinkObj()
3971          * Returns an array of links found, indexed by PDBK:
3972          *  0 - broken
3973          *  1 - normal link
3974          *  2 - stub
3975          * $options is a bit field, RLH_FOR_UPDATE to select for update
3976          */
3977         function replaceLinkHolders( &$text, $options = 0 ) {
3978                 global $wgUser;
3979                 global $wgContLang;
3980
3981                 $fname = 'Parser::replaceLinkHolders';
3982                 wfProfileIn( $fname );
3983
3984                 $pdbks = array();
3985                 $colours = array();
3986                 $sk = $this->mOptions->getSkin();
3987                 $linkCache =& LinkCache::singleton();
3988
3989                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3990                         wfProfileIn( $fname.'-check' );
3991                         $dbr = wfGetDB( DB_SLAVE );
3992                         $page = $dbr->tableName( 'page' );
3993                         $threshold = $wgUser->getOption('stubthreshold');
3994
3995                         # Sort by namespace
3996                         asort( $this->mLinkHolders['namespaces'] );
3997
3998                         # Generate query
3999                         $query = false;
4000                         $current = null;
4001                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4002                                 # Make title object
4003                                 $title = $this->mLinkHolders['titles'][$key];
4004
4005                                 # Skip invalid entries.
4006                                 # Result will be ugly, but prevents crash.
4007                                 if ( is_null( $title ) ) {
4008                                         continue;
4009                                 }
4010                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4011
4012                                 # Check if it's a static known link, e.g. interwiki
4013                                 if ( $title->isAlwaysKnown() ) {
4014                                         $colours[$pdbk] = 1;
4015                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4016                                         $colours[$pdbk] = 1;
4017                                         $this->mOutput->addLink( $title, $id );
4018                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4019                                         $colours[$pdbk] = 0;
4020                                 } else {
4021                                         # Not in the link cache, add it to the query
4022                                         if ( !isset( $current ) ) {
4023                                                 $current = $ns;
4024                                                 $query =  "SELECT page_id, page_namespace, page_title";
4025                                                 if ( $threshold > 0 ) {
4026                                                         $query .= ', page_len, page_is_redirect';
4027                                                 }
4028                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4029                                         } elseif ( $current != $ns ) {
4030                                                 $current = $ns;
4031                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4032                                         } else {
4033                                                 $query .= ', ';
4034                                         }
4035
4036                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4037                                 }
4038                         }
4039                         if ( $query ) {
4040                                 $query .= '))';
4041                                 if ( $options & RLH_FOR_UPDATE ) {
4042                                         $query .= ' FOR UPDATE';
4043                                 }
4044
4045                                 $res = $dbr->query( $query, $fname );
4046
4047                                 # Fetch data and form into an associative array
4048                                 # non-existent = broken
4049                                 # 1 = known
4050                                 # 2 = stub
4051                                 while ( $s = $dbr->fetchObject($res) ) {
4052                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4053                                         $pdbk = $title->getPrefixedDBkey();
4054                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4055                                         $this->mOutput->addLink( $title, $s->page_id );
4056
4057                                         if ( $threshold >  0 ) {
4058                                                 $size = $s->page_len;
4059                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4060                                                         $colours[$pdbk] = 1;
4061                                                 } else {
4062                                                         $colours[$pdbk] = 2;
4063                                                 }
4064                                         } else {
4065                                                 $colours[$pdbk] = 1;
4066                                         }
4067                                 }
4068                         }
4069                         wfProfileOut( $fname.'-check' );
4070
4071                         # Do a second query for different language variants of links and categories
4072                         if($wgContLang->hasVariants()){
4073                                 $linkBatch = new LinkBatch();
4074                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4075                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4076                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4077
4078                                 $categories = $this->mOutput->getCategoryLinks();
4079
4080                                 // Add variants of links to link batch
4081                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4082                                         $title = $this->mLinkHolders['titles'][$key];
4083                                         if ( is_null( $title ) )
4084                                                 continue;
4085
4086                                         $pdbk = $title->getPrefixedDBkey();
4087                                         $titleText = $title->getText();
4088
4089                                         // generate all variants of the link title text
4090                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4091
4092                                         // if link was not found (in first query), add all variants to query
4093                                         if ( !isset($colours[$pdbk]) ){
4094                                                 foreach($allTextVariants as $textVariant){
4095                                                         if($textVariant != $titleText){
4096                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4097                                                                 if(is_null($variantTitle)) continue;
4098                                                                 $linkBatch->addObj( $variantTitle );
4099                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4100                                                         }
4101                                                 }
4102                                         }
4103                                 }
4104
4105                                 // process categories, check if a category exists in some variant
4106                                 foreach( $categories as $category){
4107                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4108                                         foreach($variants as $variant){
4109                                                 if($variant != $category){
4110                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4111                                                         if(is_null($variantTitle)) continue;
4112                                                         $linkBatch->addObj( $variantTitle );
4113                                                         $categoryMap[$variant] = $category;
4114                                                 }
4115                                         }
4116                                 }
4117
4118
4119                                 if(!$linkBatch->isEmpty()){
4120                                         // construct query
4121                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4122
4123                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4124                                         if ( $threshold > 0 ) {
4125                                                 $variantQuery .= ', page_len, page_is_redirect';
4126                                         }
4127
4128                                         $variantQuery .= " FROM $page WHERE $titleClause";
4129                                         if ( $options & RLH_FOR_UPDATE ) {
4130                                                 $variantQuery .= ' FOR UPDATE';
4131                                         }
4132
4133                                         $varRes = $dbr->query( $variantQuery, $fname );
4134
4135                                         // for each found variants, figure out link holders and replace
4136                                         while ( $s = $dbr->fetchObject($varRes) ) {
4137
4138                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4139                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4140                                                 $vardbk = $variantTitle->getDBkey();
4141
4142                                                 $holderKeys = array();
4143                                                 if(isset($variantMap[$varPdbk])){
4144                                                         $holderKeys = $variantMap[$varPdbk];
4145                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4146                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4147                                                 }
4148
4149                                                 // loop over link holders
4150                                                 foreach($holderKeys as $key){
4151                                                         $title = $this->mLinkHolders['titles'][$key];
4152                                                         if ( is_null( $title ) ) continue;
4153
4154                                                         $pdbk = $title->getPrefixedDBkey();
4155
4156                                                         if(!isset($colours[$pdbk])){
4157                                                                 // found link in some of the variants, replace the link holder data
4158                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4159                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4160
4161                                                                 // set pdbk and colour
4162                                                                 $pdbks[$key] = $varPdbk;
4163                                                                 if ( $threshold >  0 ) {
4164                                                                         $size = $s->page_len;
4165                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4166                                                                                 $colours[$varPdbk] = 1;
4167                                                                         } else {
4168                                                                                 $colours[$varPdbk] = 2;
4169                                                                         }
4170                                                                 }
4171                                                                 else {
4172                                                                         $colours[$varPdbk] = 1;
4173                                                                 }
4174                                                         }
4175                                                 }
4176
4177                                                 // check if the object is a variant of a category
4178                                                 if(isset($categoryMap[$vardbk])){
4179                                                         $oldkey = $categoryMap[$vardbk];
4180                                                         if($oldkey != $vardbk)
4181                                                                 $varCategories[$oldkey]=$vardbk;
4182                                                 }
4183                                         }
4184
4185                                         // rebuild the categories in original order (if there are replacements)
4186                                         if(count($varCategories)>0){
4187                                                 $newCats = array();
4188                                                 $originalCats = $this->mOutput->getCategories();
4189                                                 foreach($originalCats as $cat => $sortkey){
4190                                                         // make the replacement
4191                                                         if( array_key_exists($cat,$varCategories) )
4192                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4193                                                         else $newCats[$cat] = $sortkey;
4194                                                 }
4195                                                 $this->mOutput->setCategoryLinks($newCats);
4196                                         }
4197                                 }
4198                         }
4199
4200                         # Construct search and replace arrays
4201                         wfProfileIn( $fname.'-construct' );
4202                         $replacePairs = array();
4203                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4204                                 $pdbk = $pdbks[$key];
4205                                 $searchkey = "<!--LINK $key-->";
4206                                 $title = $this->mLinkHolders['titles'][$key];
4207                                 if ( empty( $colours[$pdbk] ) ) {
4208                                         $linkCache->addBadLinkObj( $title );
4209                                         $colours[$pdbk] = 0;
4210                                         $this->mOutput->addLink( $title, 0 );
4211                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4212                                                                         $this->mLinkHolders['texts'][$key],
4213                                                                         $this->mLinkHolders['queries'][$key] );
4214                                 } elseif ( $colours[$pdbk] == 1 ) {
4215                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4216                                                                         $this->mLinkHolders['texts'][$key],
4217                                                                         $this->mLinkHolders['queries'][$key] );
4218                                 } elseif ( $colours[$pdbk] == 2 ) {
4219                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4220                                                                         $this->mLinkHolders['texts'][$key],
4221                                                                         $this->mLinkHolders['queries'][$key] );
4222                                 }
4223                         }
4224                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4225                         wfProfileOut( $fname.'-construct' );
4226
4227                         # Do the thing
4228                         wfProfileIn( $fname.'-replace' );
4229                         $text = preg_replace_callback(
4230                                 '/(<!--LINK .*?-->)/',
4231                                 $replacer->cb(),
4232                                 $text);
4233
4234                         wfProfileOut( $fname.'-replace' );
4235                 }
4236
4237                 # Now process interwiki link holders
4238                 # This is quite a bit simpler than internal links
4239                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4240                         wfProfileIn( $fname.'-interwiki' );
4241                         # Make interwiki link HTML
4242                         $replacePairs = array();
4243                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4244                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4245                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4246                         }
4247                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4248
4249                         $text = preg_replace_callback(
4250                                 '/<!--IWLINK (.*?)-->/',
4251                                 $replacer->cb(),
4252                                 $text );
4253                         wfProfileOut( $fname.'-interwiki' );
4254                 }
4255
4256                 wfProfileOut( $fname );
4257                 return $colours;
4258         }
4259
4260         /**
4261          * Replace <!--LINK--> link placeholders with plain text of links
4262          * (not HTML-formatted).
4263          * @param string $text
4264          * @return string
4265          */
4266         function replaceLinkHoldersText( $text ) {
4267                 $fname = 'Parser::replaceLinkHoldersText';
4268                 wfProfileIn( $fname );
4269
4270                 $text = preg_replace_callback(
4271                         '/<!--(LINK|IWLINK) (.*?)-->/',
4272                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4273                         $text );
4274
4275                 wfProfileOut( $fname );
4276                 return $text;
4277         }
4278
4279         /**
4280          * @param array $matches
4281          * @return string
4282          * @private
4283          */
4284         function replaceLinkHoldersTextCallback( $matches ) {
4285                 $type = $matches[1];
4286                 $key  = $matches[2];
4287                 if( $type == 'LINK' ) {
4288                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4289                                 return $this->mLinkHolders['texts'][$key];
4290                         }
4291                 } elseif( $type == 'IWLINK' ) {
4292                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4293                                 return $this->mInterwikiLinkHolders['texts'][$key];
4294                         }
4295                 }
4296                 return $matches[0];
4297         }
4298
4299         /**
4300          * Tag hook handler for 'pre'.
4301          */
4302         function renderPreTag( $text, $attribs ) {
4303                 // Backwards-compatibility hack
4304                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4305
4306                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4307                 return wfOpenElement( 'pre', $attribs ) .
4308                         Xml::escapeTagsOnly( $content ) .
4309                         '</pre>';
4310         }
4311
4312         /**
4313          * Renders an image gallery from a text with one line per image.
4314          * text labels may be given by using |-style alternative text. E.g.
4315          *   Image:one.jpg|The number "1"
4316          *   Image:tree.jpg|A tree
4317          * given as text will return the HTML of a gallery with two images,
4318          * labeled 'The number "1"' and
4319          * 'A tree'.
4320          */
4321         function renderImageGallery( $text, $params ) {
4322                 $ig = new ImageGallery();
4323                 $ig->setContextTitle( $this->mTitle );
4324                 $ig->setShowBytes( false );
4325                 $ig->setShowFilename( false );
4326                 $ig->setParsing();
4327                 $ig->useSkin( $this->mOptions->getSkin() );
4328
4329                 if( isset( $params['caption'] ) ) {
4330                         $caption = $params['caption'];
4331                         $caption = htmlspecialchars( $caption );
4332                         $caption = $this->replaceInternalLinks( $caption );
4333                         $ig->setCaptionHtml( $caption );
4334                 }
4335                 if( isset( $params['perrow'] ) ) {
4336                         $ig->setPerRow( $params['perrow'] );
4337                 }
4338                 if( isset( $params['widths'] ) ) {
4339                         $ig->setWidths( $params['widths'] );
4340                 }
4341                 if( isset( $params['heights'] ) ) {
4342                         $ig->setHeights( $params['heights'] );
4343                 }
4344
4345                 $lines = explode( "\n", $text );
4346                 foreach ( $lines as $line ) {
4347                         # match lines like these:
4348                         # Image:someimage.jpg|This is some image
4349                         $matches = array();
4350                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4351                         # Skip empty lines
4352                         if ( count( $matches ) == 0 ) {
4353                                 continue;
4354                         }
4355                         $tp = Title::newFromText( $matches[1] );
4356                         $nt =& $tp;
4357                         if( is_null( $nt ) ) {
4358                                 # Bogus title. Ignore these so we don't bomb out later.
4359                                 continue;
4360                         }
4361                         if ( isset( $matches[3] ) ) {
4362                                 $label = $matches[3];
4363                         } else {
4364                                 $label = '';
4365                         }
4366
4367                         $pout = $this->parse( $label,
4368                                 $this->mTitle,
4369                                 $this->mOptions,
4370                                 false, // Strip whitespace...?
4371                                 false  // Don't clear state!
4372                         );
4373                         $html = $pout->getText();
4374
4375                         $ig->add( new Image( $nt ), $html );
4376
4377                         # Only add real images (bug #5586)
4378                         if ( $nt->getNamespace() == NS_IMAGE ) {
4379                                 $this->mOutput->addImage( $nt->getDBkey() );
4380                         }
4381                 }
4382                 return $ig->toHTML();
4383         }
4384
4385         /**
4386          * Parse image options text and use it to make an image
4387          */
4388         function makeImage( $nt, $options ) {
4389                 global $wgUseImageResize, $wgDjvuRenderer;
4390
4391                 # Check if the options text is of the form "options|alt text"
4392                 # Options are:
4393                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4394                 #  * left               no resizing, just left align. label is used for alt= only
4395                 #  * right              same, but right aligned
4396                 #  * none               same, but not aligned
4397                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4398                 #  * center             center the image
4399                 #  * framed             Keep original image size, no magnify-button.
4400                 # vertical-align values (no % or length right now):
4401                 #  * baseline
4402                 #  * sub
4403                 #  * super
4404                 #  * top
4405                 #  * text-top
4406                 #  * middle
4407                 #  * bottom
4408                 #  * text-bottom
4409
4410                 $part = array_map( 'trim', explode( '|', $options) );
4411
4412                 $mwAlign = array();
4413                 $alignments = array( 'left', 'right', 'center', 'none', 'baseline', 'sub', 'super', 'top', 'text-top', 'middle', 'bottom', 'text-bottom' );
4414                 foreach ( $alignments as $alignment ) {
4415                         $mwAlign[$alignment] =& MagicWord::get( 'img_'.$alignment );
4416                 }
4417                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4418                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4419                 $mwWidth  =& MagicWord::get( 'img_width' );
4420                 $mwFramed =& MagicWord::get( 'img_framed' );
4421                 $mwPage   =& MagicWord::get( 'img_page' );
4422                 $caption = '';
4423
4424                 $width = $height = $framed = $thumb = false;
4425                 $page = null;
4426                 $manual_thumb = '' ;
4427                 $align = $valign = '';
4428
4429                 foreach( $part as $val ) {
4430                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4431                                 $thumb=true;
4432                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4433                                 # use manually specified thumbnail
4434                                 $thumb=true;
4435                                 $manual_thumb = $match;
4436                         } else {
4437                                 foreach( $alignments as $alignment ) {
4438                                         if ( ! is_null( $mwAlign[$alignment]->matchVariableStartToEnd($val) ) ) {
4439                                                 switch ( $alignment ) {
4440                                                         case 'left': case 'right': case 'center': case 'none':
4441                                                                 $align = $alignment; break;
4442                                                         default:
4443                                                                 $valign = $alignment;
4444                                                 }
4445                                                 continue 2;
4446                                         }
4447                                 }
4448                                 if ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4449                                 && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4450                                         # Select a page in a multipage document
4451                                         $page = $match;
4452                                 } elseif ( $wgUseImageResize && !$width && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4453                                         wfDebug( "img_width match: $match\n" );
4454                                         # $match is the image width in pixels
4455                                         $m = array();
4456                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4457                                                 $width = intval( $m[1] );
4458                                                 $height = intval( $m[2] );
4459                                         } else {
4460                                                 $width = intval($match);
4461                                         }
4462                                 } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4463                                         $framed=true;
4464                                 } else {
4465                                         $caption = $val;
4466                                 }
4467                         }
4468                 }
4469                 # Strip bad stuff out of the alt text
4470                 $alt = $this->replaceLinkHoldersText( $caption );
4471
4472                 # make sure there are no placeholders in thumbnail attributes
4473                 # that are later expanded to html- so expand them now and
4474                 # remove the tags
4475                 $alt = $this->mStripState->unstripBoth( $alt );
4476                 $alt = Sanitizer::stripAllTags( $alt );
4477
4478                 # Linker does the rest
4479                 $sk = $this->mOptions->getSkin();
4480                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page, $valign );
4481         }
4482
4483         /**
4484          * Set a flag in the output object indicating that the content is dynamic and
4485          * shouldn't be cached.
4486          */
4487         function disableCache() {
4488                 wfDebug( "Parser output marked as uncacheable.\n" );
4489                 $this->mOutput->mCacheTime = -1;
4490         }
4491
4492         /**#@+
4493          * Callback from the Sanitizer for expanding items found in HTML attribute
4494          * values, so they can be safely tested and escaped.
4495          * @param string $text
4496          * @param array $args
4497          * @return string
4498          * @private
4499          */
4500         function attributeStripCallback( &$text, $args ) {
4501                 $text = $this->replaceVariables( $text, $args );
4502                 $text = $this->mStripState->unstripBoth( $text );
4503                 return $text;
4504         }
4505
4506         /**#@-*/
4507
4508         /**#@+
4509          * Accessor/mutator
4510          */
4511         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4512         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4513         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4514         /**#@-*/
4515
4516         /**#@+
4517          * Accessor
4518          */
4519         function getTags() { return array_keys( $this->mTagHooks ); }
4520         /**#@-*/
4521
4522
4523         /**
4524          * Break wikitext input into sections, and either pull or replace
4525          * some particular section's text.
4526          *
4527          * External callers should use the getSection and replaceSection methods.
4528          *
4529          * @param $text Page wikitext
4530          * @param $section Numbered section. 0 pulls the text before the first
4531          *                 heading; other numbers will pull the given section
4532          *                 along with its lower-level subsections.
4533          * @param $mode One of "get" or "replace"
4534          * @param $newtext Replacement text for section data.
4535          * @return string for "get", the extracted section text.
4536          *                for "replace", the whole page with the section replaced.
4537          */
4538         private function extractSections( $text, $section, $mode, $newtext='' ) {
4539                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4540                 # comments to be stripped as well)
4541                 $stripState = new StripState;
4542
4543                 $oldOutputType = $this->mOutputType;
4544                 $oldOptions = $this->mOptions;
4545                 $this->mOptions = new ParserOptions();
4546                 $this->setOutputType( OT_WIKI );
4547
4548                 $striptext = $this->strip( $text, $stripState, true );
4549
4550                 $this->setOutputType( $oldOutputType );
4551                 $this->mOptions = $oldOptions;
4552
4553                 # now that we can be sure that no pseudo-sections are in the source,
4554                 # split it up by section
4555                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4556                 $comment = "(?:$uniq-!--.*?QINU)";
4557                 $secs = preg_split(
4558                         "/
4559                         (
4560                                 ^
4561                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4562                                 (=+) # Should this be limited to 6?
4563                                 .+?  # Section title...
4564                                 \\2  # Ending = count must match start
4565                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4566                                 $
4567                         |
4568                                 <h([1-6])\b.*?>
4569                                 .*?
4570                                 <\/h\\3\s*>
4571                         )
4572                         /mix",
4573                         $striptext, -1,
4574                         PREG_SPLIT_DELIM_CAPTURE);
4575
4576                 if( $mode == "get" ) {
4577                         if( $section == 0 ) {
4578                                 // "Section 0" returns the content before any other section.
4579                                 $rv = $secs[0];
4580                         } else {
4581                                 //track missing section, will replace if found.
4582                                 $rv = $newtext;
4583                         }
4584                 } elseif( $mode == "replace" ) {
4585                         if( $section == 0 ) {
4586                                 $rv = $newtext . "\n\n";
4587                                 $remainder = true;
4588                         } else {
4589                                 $rv = $secs[0];
4590                                 $remainder = false;
4591                         }
4592                 }
4593                 $count = 0;
4594                 $sectionLevel = 0;
4595                 for( $index = 1; $index < count( $secs ); ) {
4596                         $headerLine = $secs[$index++];
4597                         if( $secs[$index] ) {
4598                                 // A wiki header
4599                                 $headerLevel = strlen( $secs[$index++] );
4600                         } else {
4601                                 // An HTML header
4602                                 $index++;
4603                                 $headerLevel = intval( $secs[$index++] );
4604                         }
4605                         $content = $secs[$index++];
4606
4607                         $count++;
4608                         if( $mode == "get" ) {
4609                                 if( $count == $section ) {
4610                                         $rv = $headerLine . $content;
4611                                         $sectionLevel = $headerLevel;
4612                                 } elseif( $count > $section ) {
4613                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4614                                                 $rv .= $headerLine . $content;
4615                                         } else {
4616                                                 // Broke out to a higher-level section
4617                                                 break;
4618                                         }
4619                                 }
4620                         } elseif( $mode == "replace" ) {
4621                                 if( $count < $section ) {
4622                                         $rv .= $headerLine . $content;
4623                                 } elseif( $count == $section ) {
4624                                         $rv .= $newtext . "\n\n";
4625                                         $sectionLevel = $headerLevel;
4626                                 } elseif( $count > $section ) {
4627                                         if( $headerLevel <= $sectionLevel ) {
4628                                                 // Passed the section's sub-parts.
4629                                                 $remainder = true;
4630                                         }
4631                                         if( $remainder ) {
4632                                                 $rv .= $headerLine . $content;
4633                                         }
4634                                 }
4635                         }
4636                 }
4637                 if (is_string($rv))
4638                         # reinsert stripped tags
4639                         $rv = trim( $stripState->unstripBoth( $rv ) );
4640
4641                 return $rv;
4642         }
4643
4644         /**
4645          * This function returns the text of a section, specified by a number ($section).
4646          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4647          * the first section before any such heading (section 0).
4648          *
4649          * If a section contains subsections, these are also returned.
4650          *
4651          * @param $text String: text to look in
4652          * @param $section Integer: section number
4653          * @param $deftext: default to return if section is not found
4654          * @return string text of the requested section
4655          */
4656         public function getSection( $text, $section, $deftext='' ) {
4657                 return $this->extractSections( $text, $section, "get", $deftext );
4658         }
4659
4660         public function replaceSection( $oldtext, $section, $text ) {
4661                 return $this->extractSections( $oldtext, $section, "replace", $text );
4662         }
4663
4664         /**
4665          * Get the timestamp associated with the current revision, adjusted for
4666          * the default server-local timestamp
4667          */
4668         function getRevisionTimestamp() {
4669                 if ( is_null( $this->mRevisionTimestamp ) ) {
4670                         wfProfileIn( __METHOD__ );
4671                         global $wgContLang;
4672                         $dbr = wfGetDB( DB_SLAVE );
4673                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4674                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4675
4676                         // Normalize timestamp to internal MW format for timezone processing.
4677                         // This has the added side-effect of replacing a null value with
4678                         // the current time, which gives us more sensible behavior for
4679                         // previews.
4680                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4681
4682                         // The cryptic '' timezone parameter tells to use the site-default
4683                         // timezone offset instead of the user settings.
4684                         //
4685                         // Since this value will be saved into the parser cache, served
4686                         // to other users, and potentially even used inside links and such,
4687                         // it needs to be consistent for all visitors.
4688                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4689
4690                         wfProfileOut( __METHOD__ );
4691                 }
4692                 return $this->mRevisionTimestamp;
4693         }
4694
4695         /**
4696          * Mutator for $mDefaultSort
4697          *
4698          * @param $sort New value
4699          */
4700         public function setDefaultSort( $sort ) {
4701                 $this->mDefaultSort = $sort;
4702         }
4703
4704         /**
4705          * Accessor for $mDefaultSort
4706          * Will use the title/prefixed title if none is set
4707          *
4708          * @return string
4709          */
4710         public function getDefaultSort() {
4711                 if( $this->mDefaultSort !== false ) {
4712                         return $this->mDefaultSort;
4713                 } else {
4714                         return $this->mTitle->getNamespace() == NS_CATEGORY
4715                                         ? $this->mTitle->getText()
4716                                         : $this->mTitle->getPrefixedText();
4717                 }
4718         }
4719
4720 }
4721
4722 /**
4723  * @todo document, briefly.
4724  */
4725 class OnlyIncludeReplacer {
4726         var $output = '';
4727
4728         function replace( $matches ) {
4729                 if ( substr( $matches[1], -1 ) == "\n" ) {
4730                         $this->output .= substr( $matches[1], 0, -1 );
4731                 } else {
4732                         $this->output .= $matches[1];
4733                 }
4734         }
4735 }
4736
4737 /**
4738  * @todo document, briefly.
4739  */
4740 class StripState {
4741         var $general, $nowiki;
4742
4743         function __construct() {
4744                 $this->general = new ReplacementArray;
4745                 $this->nowiki = new ReplacementArray;
4746         }
4747
4748         function unstripGeneral( $text ) {
4749                 wfProfileIn( __METHOD__ );
4750                 $text = $this->general->replace( $text );
4751                 wfProfileOut( __METHOD__ );
4752                 return $text;
4753         }
4754
4755         function unstripNoWiki( $text ) {
4756                 wfProfileIn( __METHOD__ );
4757                 $text = $this->nowiki->replace( $text );
4758                 wfProfileOut( __METHOD__ );
4759                 return $text;
4760         }
4761
4762         function unstripBoth( $text ) {
4763                 wfProfileIn( __METHOD__ );
4764                 $text = $this->general->replace( $text );
4765                 $text = $this->nowiki->replace( $text );
4766                 wfProfileOut( __METHOD__ );
4767                 return $text;
4768         }
4769 }
4770
4771 ?>